Project 1

Remote sensing system Active Safety MMF320

Daniel Karlsson 840519-5598 Goran Kotur 840810-4613 Sabino Paolo Luisi 830924C217 François Niederlander 860103C274 Daniel Tidholm 810921-5031 Antonio Turturro 830212C150

i

Summary This report deals with the technologies of active safety in general and specifically remote sensing. Active safety systems that are on the market today have bin briefly described to get an overview of the market and also understand how they can be integrated with the remote sensing systems. For example ESC and ABS are brought up. Radar, laser and camera and also some variants of these have bin investigated and described as well as the systems they are a part of. Advantages and disadvantages for each system have bin brought up to make the comparison between them easier. For the comparison three accident types have also been used. These have been chosen after investigation of current statistics over fatalities in different accident types. Problems with active safety are discussed through the report and the human related problems are considered under the part about HMI (human machine interface). Today it has not been proven that all systems on the market really saves lives. Therefore the systems investigated are critically checked and some conclusions about there effectively are drawn. It is made clear that the autonomous systems, taking control of the car have legal problems both if the malfunction and doesn’t prevent an accident and if the misinterpret a situation and therefore cause an accident. On the other hand the informing systems warning the driver with signals transfer the decision making on the driver which also gives some problems e.g. information overload. The recommendation to the board of directors is in the short term to manufacture a system with both short and long range radar for vehicle and pedestrian detection in the situations considered. In the long term it is suitable to further investigate the possibility of making a system using camera and radar with powerful image processing for even better performance at a reasonable cost.

ii

Table of contents 0 Introduction ............................................................................................................................. 1

0.1 Background ...................................................................................................................... 1 0.2 Description of the task...................................................................................................... 1 0.3 Goal .................................................................................................................................. 1 0.4 Aim................................................................................................................................... 1 0.5 Focus ................................................................................................................................ 2 0.6 Delimitation...................................................................................................................... 2 0.7 Methods............................................................................................................................ 2

1. Active Safety .......................................................................................................................... 3 1.1 Active Safety, history and potential ................................................................................. 3 1.2 Anti-lock braking system (ABS)...................................................................................... 4 1.3 Dynamic Brake Control (DBC)........................................................................................ 4 1.4 Electronic Brake force Distribution (EBD)...................................................................... 5 1.5 Electronic Stability Control (ESC)................................................................................... 6 1.6 Traction Control System (TCS) ....................................................................................... 8 1.7 Active Steering................................................................................................................. 9 1.8 Air Bag System ................................................................................................................ 9

2. Remote Sensing Systems ..................................................................................................... 10 2.1 Camera Vision................................................................................................................ 10

2.1.1 Design for automotive applications......................................................................... 10 2.1.2 Stereovision systems ............................................................................................... 11 2.1.3 Market situation....................................................................................................... 11 2.1.4 Lane Departure Warning System (LDWS) ............................................................. 12 2.1.5 Driver Seering Recommendation (DSR)................................................................. 13 2.1.6 Blind Spot Information System (BLIS) .................................................................. 13 2.1.7 Different vision systems and providers................................................................... 15

2.2 Night Vision [15,16 ] ..................................................................................................... 16 2.2.1 Comparison between BMW and Mercedes night vision systems ........................... 16 2.2.2 Future development................................................................................................. 19

2.3 Radar Vision................................................................................................................... 19 2.3.1 Requirements........................................................................................................... 20 2.3.2 Problems.................................................................................................................. 20 2.3.3 Analysis of Systems ................................................................................................ 21 2.3.4 Active Cruise Control.............................................................................................. 21 2.3.5 Adaptive Cruise Control with Stop & Go ............................................................... 23 2.3.6 Collision Warning ................................................................................................... 24 2.3.7 Collision Warning and Preparation ......................................................................... 24 2.3.8 Pre-Crash Emergency Braking with Millimeter Wave Radar and Stereo Camera Fusion System .................................................................................................................. 24 2.3.9 Rear Pre Crash Safety System................................................................................. 25 2.3.10 Status Of Radar Based Automotive Applications ................................................. 25 2.3.11 76-77 GHz Long Range Radar Systems ............................................................... 25 2.3.12 24 GHz Short Range Radar................................................................................... 26 2.3.13 Future Radar Developments.................................................................................. 26

2.4 Automotive Laser Scanner ............................................................................................. 27 2.4.1 Signal processing..................................................................................................... 29

2.5 Ultrasonic Sensors.......................................................................................................... 30

iii

2.6 Example of Fully integrated cars ................................................................................... 31 2.6.1 The Springrobot....................................................................................................... 31 2.6.2 Mitsubishi Active Safety ASV................................................................................ 32

2.7 Integration of active safety and remote sensing ............................................................. 34 3. Competing technologies....................................................................................................... 35 4. HMI ...................................................................................................................................... 37

4.1 Autonomous and informing systems.............................................................................. 37 4.2 Warning signals and information systems ..................................................................... 38 4.3 Driver performance and user friendliness ...................................................................... 38

5. Analysis of traffic incident types ......................................................................................... 39 5.1 What does that statistics say? ......................................................................................... 39 5.2 Why do accidents occur?................................................................................................ 40 5.3 Three accident types....................................................................................................... 41

5.3.1 Frontal collision in intersection............................................................................... 41 5.3.2 Proposed solution for frontal collision in intersection ............................................ 41 5.3.3 Rear end collision.................................................................................................... 42 5.3.4 Proposed solution for rear end collision.................................................................. 42 5.3.5 Pedestrian accidents ................................................................................................ 42 5.3.6 Proposed solution for pedestrian accidents ............................................................. 43 5.3.7 Combined proposition ............................................................................................. 43

6. Manufacturing Strategies ..................................................................................................... 43 6.1 Market Analysis ............................................................................................................. 44 6.2 Potential Fields for Production....................................................................................... 44 6.3 The Future ...................................................................................................................... 44

7. Discussion ............................................................................................................................ 45 Refrences.................................................................................................................................. 46 Appendix A ..................................................................................................................................

1

0 Introduction Cars of today are getting ever more sophisticated in terms of performance and safety. Especially safety has come to play a more central role, and over the last decade the field of active safety has become somewhat of the Holy Grail in vehicle safety development. Car safety is made up of two subfields namely active safety and passive safety. Active safety is a term used for systems made up of sensors, modern computer equipment and software. These systems have the purpose of aiding the driver in dangerous situations and keep maneuverability of the vehicle to avoid accidents. Typical traditional active safety systems are Anti-lock Brakes and Stability Control Programs and these can be found installed on large portion of the modern car park. However, there are a lot of new exotic systems appearing on the market that uses sensing modern sensing technology. These systems will be analyzed in further detail in the report. Passive safety contains all features that are designed to protect the occupants in the vehicle if an accident actually occurs. Typical examples are airbags, seatbelt, inflatable curtains and reinforcements in the chassis. Active safety systems have been around for quite a while now. However, there has come a relatively new set of technologies to the active safety field, and that is something called “remote sensing”. Remote sensing is a group of systems that use radar, laser, cameras, lidar and ultra waves to gather information about the surrounding of the car. It is implemented in systems like Lane Departure Warning, Adaptive Cruise Control and Night vision. Most of these remote sensing systems have not yet made it into the mainstream of vehicles but are mainly available as options in higher end cars.

0.1 Background A company producing remote sensing systems is about to expand. The board of directors will soon decide the outlines for a large scale project with the aim of upgrading the product line. One opportunity is to start making sensors for the automotive industry, used in systems for active safety. To be able to make a correct decision, the board of directors has ordered us to form a task force with commission to address all potential of remote sensing systems in vehicles for improved traffic safety and deliver the result in a report.

0.2 Description of the task The task is to describe the remote sensing systems available on the market and compare this system to each other as well as to other safety systems. Competing technologies will be briefly described to make the choice of system easier. Further development of remote sensors will be discussed and in case of breakthrough a new system will be proposed. All systems will be analyzed thro unbiased and subjective comparison of their performance in different driving situations. Driving situations and accidents of interest will be addressed.

0.3 Goal The goal is to hand over a report to the board of directors that in detail describes the potentials and limitations with remote sensing systems. The systems that the sensors are a part of should be investigated to enable the directors to make a good decision for the company’s future investments.

0.4 Aim To propose remote sensors well suited to remote sensing systems that may prevent accidents or reduce the number of injured and killed people in accidents. The report shall summaries the

2

state of the art in technical solutions that are being developed to deal with this type of safety systems.

0.5 Focus The focus is on investigating remote sensing systems and the systems they are a part of. How this systems work and how the prevent accidents will be described. The possibilities and problems with the safety systems interaction with other systems on the car and with the driver will also be discussed.

0.6 Delimitation Other safety systems that may not compete with systems of remote sensing want be considered. In lack of more exact information, the price of the products will only be an estimation and comparison between the different techniques.

0.7 Methods First the task will be analyzed to address the problem and way of work. Existing systems and sensing techniques will be analyzed as well as other active safety systems that may compete on the market. This will be done by research of technical reports and information from car manufacturers. It is important to remember that some information provided by manufacturers have the purpose to sell products and should therefore be questioned. Different signal systems that inform the driver will be described and human machine interface discussed. The next step is to find traffic situations and accident types that are interesting for active safety and remote sensing. These will be investigated and chosen through statistical inquiry and subjective opinion. After which the situations will be used to compare the systems and propose combinations of them or other solutions to prevent accidents. Through a simple version of the ((choosing process)) a system will be designed. With help from the information brought through the work described above and discussion of the availability and other parameters a new system will be developed. Finally the solution will be recommended to the board of directors.

3

1. Active Safety Much of the car industry’s resources today are being focused on developing safety systems to avoid and prevent accidents, reduce injuries and save lives. Some companies want to strengthen their brand and image with good safety, because safety is a strong sell argument. Today, nearly all car manufactures have some sort of active safety equipment in their cars. The term “Active Safety” incorporates many different technologies such as brakes, steering and suspensions. These safety systems have the purpose of enhancing vehicle control, avoid accidents as far as possible, and help to make the driver experience safer. Together with recent development of more advanced computers and software, more sophisticated safety systems can be integrated into the car, such as stability control and roll over protection. Traditionally, most cars have only been equipped with Anti-lock brakes (ABS), but over the last years more advanced systems have started to appear, and they have gotten a fairly good market penetration which can be seen in fig.1 below. In this section a brief introduction to various active safety systems will be described together with illustrative pictures to give the reader a good knowledge about how the different systems work.

1.1 Active Safety, history and potential The history of active safety technologies goes back roughly 30 years (see fig.2) with the introduction of ABS breaking systems in the mid 70’s. After that the development of new safety systems has been steady and today almost every new car sold is equipped with at least ABS brakes (see fig.1). Today higher-end cars feature more advanced systems such as stability control, traction control, dynamic breaking and even state of the art systems that connects many of these independent systems to an integrated system that is more capable of judging and avoiding dangerous traffic situations. Much development today is done on stability control systems, ESP, since this prevents many accidents caused by skidding, which is very dangerous. The long trends also indicates that more of these independent systems will be improved and integrated into a “vehicle dynamics integrated management”-system so called VDIM, to become more effective.

Fig.1 2003 Vehicle Safety Feature Fig.2 Timeline for Active Safety development [2] installation rates. [1]

4

1.2 Anti-lock braking system (ABS) ABS brakes are the most commonly installed active safety system in cars up to date, almost every new sold car is equipped with it.

Fig.3 Without ABS (left) and with ABS (right) [3] The ABS system is made up of a speed sensor at every wheel. These sensors continuously send signals to a computer which monitors if any of the wheel’s speed suddenly drops more than above-average. If so, the computer detects this as a risk of locking the wheel and immediately reduces the hydraulic brake pressure to the corresponding wheel so it can rotate freely. This control cycle can be repeated many times every second if necessary and the driver feel a pulsation in the pedal as a sign the ABS working. Advantages of ABS

• Short braking distances by utilizing maximum road friction • Maintenance of steering capability even under extreme braking conditions

Disadvantages

• Under some conditions like ice, packed snow, and gravel, ABS can actually lengthen stopping distances where a locked tire would build up more debris in front of the tire and stop quicker than if the tire rolls some.

• Another disadvantage is that most people never use the ABS because they never practice.

Thanks to the ABS system one can apply maximum break pressure with no risk of locking the wheels, thus being able to keep steering ability during a sudden breaking maneuver, even with varying road condition on each wheel. By not locking the breaks, skidding is greatly reduced and chances of a safe outcome are increased.

1.3 Dynamic Brake Control (DBC) In an emergency situation, 90% of car drivers are incapable of effectively execute an emergency stop. A typical response is to react quickly but not apply sufficient initial brake pressure to the brake pedal, thus not fully utilize the breaking power and valuable, life saving breaking distance is wasted. Fig.3 EBA, increased brake force is applied in case of an emergency [3]

5

To fully exploit the vehicle’s breaking power a separate safety system called “Emergency Brake Assist” (EBA) or “Dynamic Brake Control” (DBC) has been designed. The components of the system are made up of sensors in the hydraulic brake pressure system working together with a computer. When the sensors detect a rapid increase in hydraulic braking pressure at a certain high level in the main brake cylinder, the computer interprets this as an emergency stop. The computer then signals to accelerate and reinforce the build-up of brake pressure. This helps the vehicle to reach its shortest possible breaking distance if insufficient braking pressure is applied to the pedal. Advantages of DBC

• Detects emergency stop and increased braking pressure is applied • Builds up pressure until both axels are regulated by ABS • Braking power is exploited to the fullest. • Minimizes the vehicle’s braking distance, saving lives

1.4 Electronic Brake force Distribution (EBD) EBD is designed to optimize and better utilize the performance of the ABS braking system. An ordinary ABS prevents individual wheels from locking by cutting braking pressure when skidding is detected. EBD goes one step further and electronically distributes brake pressure between front and rear wheels depending on present load in the car and road conditions, e.g. passenger in the back seat as seen in fig.6 and 7. EBD can also distribute braking force on individual wheels, e.g. braking wheels on one side slightly harder than the other to maintain stability and control during a corner. [4] Fig.4 Sensors and computer [4] Fig.5 Brake Actuator [4] The function of EBD is made up of speed sensors that are located on all four wheels, ECU (see fig.4) and Brake Actuator. The speed sensors monitor if one or more wheels have locked or may lock during strong braking. If so, the sensors send signals to the ECU about the current situation. Instantaneously the computer calculates which wheels needs to be braked and which to be released to avoid skid. It then sends a signal to the Brake Actuator (fig.5) which distributes the brake pressure to all four wheels as required to enable stable vehicle control. [4]

6

Fig.6 & 7 Electronic Brake Control distributes brake force on individual wheels.[3]

1.5 Electronic Stability Control (ESC) ESC is a computer-controlled system that helps the driver retain control of the vehicle during extreme cornering or in other emergency situations. There are many different names of similar systems used in the car industry today that has the same function as ESC. The following systems, used by different manufacturers, are equivalent to ESC.

• Active Stability Control (ASC) • Dynamic Stability Control (DSC) • Electronic Stability Program (ESP) • Vehicle Dynamic Control (VDC) • Vehicle Stability Assist (VSA) • Vehicle Stability Control (VSC)

Electronic Stability Control is a further development of the Anti-lock Brake System (ABS) and Traction Control System (TCS) together with sensors. While TCS only registers and takes into account forces that act in the driving direction, the ESC system additionally registers and analyzes dynamic side forces and a lot of other parameters to ensure vehicle stability. The functionality of the ESC system is made up a several different components together with ABS.

• Steering angle sensor to detect the driver's chosen path • Lateral-acceleration sensor to define the lateral forces which must be transmitted to

the road by the tires • Rate-of-turn sensor to define the vehicle's yaw rate • Brake-pressure sensor to define the longitudinal forces acting between tires and road

surface during braking [5] The ESC computer constantly monitors data such as steering wheel movement, vehicle speed, side forces and yaw rate and then computes an ideal driving condition that is then compared with the current status. If the current status deviates too much from the ideal, the ESC immediately detects this instability that might threaten to push the car of the road causing

7

injury to the driver. Therefore, when ESC detects instability, it intervenes via the engine management and can momentarily reducing drive torque and applying brakes on wheels where the system finds it necessary to keep or restore vehicle control. All this happens in a matter of milliseconds, many times faster than the human driver could ever react and respond. In reality ESC can detect e.g. over steer and counteract it by applying brakes on the outer front wheel, creating a counter acting moment to ensure vehicle stability. This can be seen in fig.10 Advantages of ESC

• A significant reduction in the risk of skidding during sudden severe maneuvers in bends or on slippery roads.

• Optimum use of road surface friction together with TCS.

Disadvantages • People used to driving vehicles equipped with ESC might over estimate their driving

skills when placed in a vehicle without ESC, thus increasing the risk of skidding.

Fig8. Old ESP, functions and components [2] Fig.9 Next generation ESP [2]

Fig.10 Explanation of over- and under steer and how ESC works

8

Under steer occurs when the car can not follow the intended path of steering. ESC reduces engine torque and applies brakes to inner side wheels to prevent front tire skid [6].Over steer occurs when the vehicle turns too sharply into the direction of steering. Side skid of the back tires is controlled by applying the brakes to the outside wheels. ESC has in many studies shown over and over again that it greatly reduces the risk for rollover crashes, thus saving lives. Therefore, in 2006 the American organization “The National Highway Traffic Safety Administration” (NHTSA), working for “The Department of Transportation”, proposed to establish a new vehicle standard. This new standard, FMVSS no.126 [7], requires electronic stability control (ESC) systems to be installed on passenger cars, multipurpose vehicles, trucks and busses with a weight of 4 536kg or less. Based on studies and traffic data, NHTSA has estimated that installation of ESC will reduce single vehicle crashes by 34%, single vehicle crashes of Sports Utility Vehicles (SUVs) by as much as 59%, and much greater reduction of rollover crashes. It has also been estimated that 71% resp. 84% of single vehicle and SUV rollovers could be prevented. The new rule would require a 100 percent rate of installation by midyear 2012. In short, ESC is one of the most effective electronic "helping hands" for drivers invented so far and should be installed in as many cars as possible.

Fig.11 Reduction of accidents due to ESC [8]

1.6 Traction Control System (TCS) The second most common active safety system installed in modern cars is “traction control systems”. TCS and ABS together make up the core components around which the first ECS system was developed, and still they are two of the most vital components that make ESC such a powerful system. The Traction Control System (TCS) makes driving in slippery conditions easier and safer. Sensors continuously monitor the rotational speed of each individual wheel. As soon as they detect that one or more wheels are starting to slip, the system instantaneously calculates the best way to restore traction. The engine's computer control unit then adjusts the throttle accordingly and the Brake Actuator directs the braking force to the most appropriate wheel until traction returns.

9

Advantages of TCS • Always having traction increases handling of the vehicle and reduces risk of skidding

Disadvantages • People used to driving vehicles equipped with TCS might over estimate their driving

skills when placed in a vehicle without TCS, thus increasing the risk of skidding.

1.7 Active Steering This system helps in the direction stability of the vehicle. Steering angle sensors situated on the steering column sense the direction the driver wants to go and only intervenes if the car is beyond stable limits. It has to occur much faster than the driver can react. An active steering system is a complementary system for a front-steered vehicle that adds or subtracts a component to the steering signal performed by the driver. The steering signal from the driver is an angular movement on the steering wheel. The resulting steering angle is thus composed by the component performed by the driver and the component contributed by the steering system. On the market, different brands propose several solutions of active steering. This is systems of steering aid integrated in cars. Basically, the main idea is to improve stability and handling in order to improve safety and comfort. Several actuators are used to reach these objectives. The steering wheel, steering rack and their mechanical connections are also important parts. Active steering is the idea of an integrated steering support system for cars. The system has to behave like the steering on conventional cars but with additional functionality such as disturbance rejection (due to for instance wind gusts or decreased road adhesion conditions). These two sections will show two examples of technical solutions of the steering systems which are used today. The two brands which produce these systems are BMW (Active Steering) and General Motors (Quadra steer).

1.8 Air Bag System Honda recently created a new kind of driver and front passenger airbags. The innovation called "dual output inflators" adjusts the deployment force of the airbags to the severity of the crash. This brand has also introduced the first system to sense when children or small-statured adults in the front passenger seat are not in the proper position and to stop deployment of the side airbag. Honda's very first front passenger air bag in the 1980s, utilized a unique design mounted to the top of the dashboard, and deployed upward rather than directly at the passenger. The Center for Auto Safety praised this front passenger airbag design as "a blueprint on how to design better airbag systems." [9] Fig.12 Illustration of airbag deployment Fig. Fig.13 Standard Occupant Position

Detection System

10

In order to use this technology efficiently, it is useful that vehicle with front side airbags includes a standard occupant position detection system. This is most of the time done for the passenger side to stop deployment of the airbag when a child or small stature adult is not in the proper position. These kinds of body types are not so common for the driver seat. That is why only the passenger seat is concerned. This is not required by any regulation and is not even available on other automobiles yet.

2. Remote Sensing Systems In the next chapters will be described the following remote sensing technologies:

• Camera vision • Radar vision • Laser vision • Ultra sound vision

and at the end there is a example of one fully integrated system embedded in a car.

2.1 Camera Vision Vision sensors offer a 2D array of up to a million pixels with a large field of view, the range detection and the angular field of view depending upon the optics (lens and focal). This rich source of information allows for much smarter, though involved, applications, for example, classification and recognition of object. Wide-luminance range sensors are provided by CMOS imagers with nonlinear luminance response and will certainly replace current charge coupled device (CCD) imagers. In fact, CMOS devices can offer several advantages over CCD-based imagers, including enhanced functionalities with individual pixel signal processing, lower power consumption, and lower cost. Within several application fields in which the advanced signal processing capabilities of CMOS imagers are useful, the automotive one is currently the object of many research activities and offer wide market opportunities. High-temporal dynamics and a fast read out resolve fast movement. Thus, the digital image sequence is passed to an evaluation unit that performs appropriate signal processing to extract the desired output.

2.1.1 Design for automotive applications The design of vision sensors for automotive vision has to take into account specific constraints. When most of the current vision systems use full colour, accurate image reproduction, and photographic or video aspect ratios, operating over a visible light spectrum of 0.45 to 0.7 µm and realized with CCD or CMOS sensors, most automotive applications are monochrome and may involve variable aspect ratios, pixel size, and pitch. Constraints:

• Automotive imagers must collect light at very low levels, and yet still resolve objects in direct sunlight. Another constraint concerns practical resolution requirements.

• The automotive vision system will be used as sensor input for computational vision that will perform higher-level task by extracting features from the imager and its pixels information in order to perform lane recognition, passenger occupancy detection, forward vehicle, or pedestrian detection [10, 11].

• Automotive durability, at an affordable cost.

11

2.1.2 Stereovision systems The ranging capability of active sensors (radars, ladars) is an extremely rich data source. A stereo based vision system can similarly provide a direct absolute measurement of the scene [12]. Computing depth from two images is a computationally intensive task. It involves finding for every pixel in the left image the corresponding pixel in the right image. Correct corresponding pixel is defined as the pixel representing the same physical point in the scene. The distance between two corresponding pixels in image coordinates is called “disparity” and is inversely proportional to distance. Advantages:

• Stereo sensing provides a colour or monochrome image, which is exactly (inherently) registered to the depth image. This image is valuable in image analysis, either using traditional 2D methods, or methods that combine colour and depth image data.

• The operating-range functions of lens field-of view, lens separation, and image size are flexible.

• Stereo sensors have no moving parts, an advantage for reliability. Computation relates to the frame rate, which needs to be high with low latency and remains an issue for many applications. In safety applications such as airbag deployment, the 3D position of vehicle occupants must be understood to determine whether an airbag can be safely deployed — a decision that must be made within tens of milliseconds (Siemens VDO, Delphi). For vehicle tracking applications, by means of 3D information, it is easy to distinguish vehicles and shadows on the ground, vehicles and reflections on the road, or detect overlapping vehicles (Omron Corp.).

Disadvantages

• The key problem of stereovision systems is a correct calibration [13,14]. In all applications where not only recognition is important, but a correct localization in world coordinates is essential, a precise mapping between image pixels and world coordinates becomes mandatory. This correspondence may vary during system operations due to many reasons; in automotive applications, vehicle movements and drifts due to sudden vibrations may change the position and orientation of the cameras, making this mapping less reliable.

2.1.3 Market situation Automotive application using cameras started to appear on the market by the end of the 1990s. The first commercial application has been the lane departure warning.

Fig.14 Lane

departure warning system.

12

This system (see fig.14), which warns the driver when the vehicle tends to drift away from its lane, is now often used in trucks. The first application, which put some control of the vehicle, was the introduction of lane keeping assistant (Nissan Cima in 2002) where a single camera helps the driver to stay in the middle of the lane by applying a corrective torque on the steering wheel.

2.1.4 Lane Departure Warning System (LDWS) LDWS is a warning system that alerts the driver when he or she accidentally changes lane without using the indicator. One reason to this behavior could be that the driver is about to fall asleep and is about to move across the road and hit cars in the opposite direction, or is about to leave the road and hit trees etcetera. LDWS uses six sensors behind the front bumper, three on each side, to monitor the lane markings. Each sensor is equipped with an infrared light-emitting diode and a detection cell. When the sensors notice that the car is drifting across the lane markings and the indicators are not in use, a computer sends a signal to a pair of vibration devices, on each side of the driver’s seat (see fig.15).

Fig.15 sequence of work for LDWS

If the car is drifting to the right, the driver feels a vibrating signal in the right side of the seat and vice versa. The warning allows the driver to take immediate actions and steer back to the lane. Lane departures are detected by variations in the reflections from the infrared beams because of the colour change between the black asphalt and the white lane markings. The system is activated by the push on a button and can also detect yellow, red and blue lane markings which are used in some European countries (see fig.16).

Fig.16 Button for activating LDWS The drawback occurs when the lane markings are covered by snow, dirt, leaves etcetera. This makes it difficult for the sensors to detect the lane markings. Lane Departure Warning System has a couple of similar subsystems:

13

Lane Keeping Aid goes one step further, compared to the LDWS. In addition to the audible signal, the system actually steps in and helps steer the car back on course. This steering capability is relatively limited. The aim is not to take over the steering. Instead, the maneuver is usually sufficient to help the driver take action to keep the vehicle within the current lane. Lane Departure Warning System and Lane Keeping Aid are switched off when speed is below 43 mph so they do not interfere in urban conditions when intentional lane changes take place all the time. The functions can also be switched off manually. A third system, Emergency Lane Assist, is a world breakthrough and is not yet available on the market. Using both camera and radar, Emergency Lane Assist can monitor oncoming vehicles. If the driver crosses the lane markers and does not respond to the buzzer, the system reacts by applying additional steering force to help steer back into the intended lane. Emergency Lane Assist is active at all speeds.

2.1.5 Driver Seering Recommendation (DSR) Driver Steering Recommendation takes advantage of the hardware of Electric Power Assisted Steering systems (EPAS). The objective of the DSR system is to recommend the driver to steer correct without forcing them; they should always be able to override the applied torque. Its purpose is to use the torque generated by its electric motor not only for easing operation on the steering wheel, but also influencing and improving the driver reaction. In comparison with Active Steering, the DSR system has some advantages

• It has a lower cost as no extra hardware is needed, provided that the car is already fitted with EPAS.

• It can actually influence in driver actions, which allows for useful collaboration with other active systems such as Lane Keeping Assistance.

Unfortunately, it has some disadvantages

• DSR has the driver in the loop and the effectiveness relies on the reaction of the driver which introduces a big uncertainty. It is hard to determine whether a surprised and unskilled driver will let go and follow the systems suggestion or interpret the change in steering torque as a natural reaction of the car which has to be overcome in order to impose their will.

• As a consequence of the latter, DSR systems are harder to tune than Active Steering systems. Margin for a great variability in reactions must be considered and a solution that will not lead to a worse driver operation on the steering wheel must be selected.

Different opinions exist if DSR is an Active Steering system or not. If the only requirement for a system to be considered an Active Steering device is to be able to influence on the steering angle by any means, DSR would be one of them. If the fact that it has to be able to do it autonomously, it wouldn’t. DSR tries to affect the driver’s inputs on the steering wheel, but the front tires will always follow their orders, as no active hardware exists on DSR systems.

2.1.6 Blind Spot Information System (BLIS) BLIS, a system developed by Volvo, warns the driver about cars that are approaching from the rear or cars that the driver is currently overtaking. The system uses a camera in each rear-view mirror and these cameras are pointed at the so called Blind Spot, meaning the area alongside of the car which is hard to monitor by the outside mirrors (see fig.17).

14

Fig.17 A camera is mounted under the outside mirror When another vehicle enters the monitored zone, a lamp comes on in the relevant mirror. The driver gets a clear indication that there is another vehicle in the risk zone and can keep away. The system provides information about cars approaching from the rear and also vehicles in front that the driver is currently overtaking. This information gives the driver added scope for taking the right decision in such situations. Both sides are monitored in the same way. The main unit behind BLIS is a digital camera which is fitted on each door mirror and takes a large number of frames a second. By comparing the picture frames, the system can register when a vehicle is moving into the monitored zone, which is 9.5 meters long and 3 meters wide (see fig.18).

Fig.18 BLIS monitors the zone which is called the blind spot.

The system is programmed to monitor cars as well as motorbikes, in both daylight and in the dark. It is also dimensioned not to react to parked cars, roadside fences, crash barriers, lampposts and so on. The system is active at all speeds above 10 km/h. It is designed to alert the driver to vehicles that are moving a maximum of 20 km/h slower and a maximum of 70 km/h faster than the driver’s own vehicle. The BLIS system costs 5 700 Swedish Crowns and can now be found in cars such as the new Volvo S80, XC90 and V70.

15

2.1.7 Different vision systems and providers Here follows a list off different vision systems and which companies provide them.

• Rearview camera which are Available from: Audi, Land Rover, Lexus, Mazda, Mercedes-Benz, Toyota and Volkswagen among others. The price varies between $750 to $1,000 — or more, if part of a package.

• Adaptive headlights and/or night-vision assist. Also available from several car

manufacturers such as BMW, Infiniti, Mercedes-Bens, Volvo and many others. Prices range in about $700 and up if part of bigger premium package.

• Lane-departure warning system comes at a cost of approximately $3,600-$10,500 from manufactures such as Citroën and others.

16

2.2 Night Vision [15,16 ] A system called Night Vision can be found in cars like BMW, Mercedes and Cadillac. Thanks to an infrared camera, mounted in the front of the car, the driver can when driving in the dark, discover a human being or an animal up to 300 meters away (see fig.19).

Fig.19 Far infra-red camera mounted under the bumper (left). Everything that generates heat such as a person, an animal and to some extent trees and bushes can easily be monitored on the display. The more heat the object generates the more clearly it is shown on the screen. This is essential since people and animal generates most heat, they are also the most important things to concentrate on when driving responsibly. Night Vision makes it possible for the driver to discover an object much sooner. When driving at the speed of 100 km/h, the driver can discover a person up to five seconds before he or she is light up by the cars headlight. This is a significant amount of time when it comes to safety. The extra five seconds helps the driver to increase the safety margins and decrease the stress. The image section also follows the road even in curves and objects far away can be enlarged (see fig.20). ¨

Fig.20 People and animal can easily be monitored on the screen.

2.2.1 Comparison between BMW and Mercedes night vision systems Night vision can be executed in different forms, such as infrared headlamps or thermal-imaging cameras, in the following chapters there is a comparison between these two different technologies:

17

• BMW: Night vision realized by AutoLiv/Flir

It uses “FIR” Far InfraRed technology (300 meters): a thermal video camera, installed in the lower part of the front bumper, records the heat irradiated by the environment and sends the data to a processor, in order to transmit it on the screen.

• MERCEDES: Night view realized by Bosch

It uses “MIR” Near InfraRed technology (200 meters), based on an infrared video camera.

The system adopted by BMW is based instead on four main features:

• The activation is realized by a switch placed to the left headlights command.

• The system works in any environmental condition with low-beam headlights activated.

• The thermal imaging camera is installed behind impact-resistant glass and a fine grid in the left front bumper.

• A CPU that elaborates the data received from the camera.

• Display Mercedes system is composed by six main components:

• Activation switch placed to the left of the lights command.

• Two special beacons embedded in the front headlight groups illuminate the road with infrared beams the system is automatically activated when the speed is higher than 15 km/h.

• Infrared camera placed on the inner side of the windshield records the scene ahead of the Fig.21 BMW and Mercedes sensor location vehicle. It is based on CMOS technology (Complementary Metal Oxide Semiconductor)

• Protection system for the camera (light reflection disturbs). • CPU that elaborates the data received from the camera.

Fig.22 BMW and Mercedes sensor location

18

Advantages and disadvantages; Image quality, range and display location BMW

• High contrast between hot and cold things, the image quality is low, but distinguishing objects is simple.

• 300 meter range. • The BMW solution offers a display on the right of the driver, as a consequence there is

fatigue, caused by the continuous head movements, and the distraction risk is higher.

Mercedes • Higher image quality, it is more difficult to distinguish objects (in fig. 4, lower part on

the right in the blue circle, there isn’t high contrast between the man and the trees). • 200 meter range. • The display location is more comfortable, because it is located in the instruments

cluster, and it is simple to check the road without moving the head. The speed is visualized in the lower part of the display (fig.4).

In both cases it is necessary to heat the protective lens in order to avoid ice formation.

Fig.23 & 24 Mercedes and BMW display location and image quality BMW cost efficiency

• Higher sensor cost • Higher system cost

Mercedes cost efficiency

• Lower sensor cost (320x240 pixel), the camera is derived from commercial cameras (cmos sensor). It is adapted by using a filter for the other wave lengths except infrared. The sensor requires an infrared headlight.

19

2.2.2 Future development To further increase the performance of night vision systems, SiemensVDO works on a new improved display interface technology. Instead of a traditional display screen in the instruments cluster, the Night Vision picture will be directly projected on the windscreen. This results in less head movements for the driver and thus gives less distraction.

Fig.25 Siemens VDO system

2.3 Radar Vision The idea to mount radar on a car isn't a new one. Since the 1960s, various ideas have been developed and tested, but the devices were either too bulky or too expensive for mass production. At the time of writing, several manufacturers around the world have developed small and lightweight devices, and efforts at making them affordable for the average driver are underway. The main advantage of radar over competing devices, such as laser or infrared vision equipment, is the radar's ability to look through rain, fog and snow. Current devices are concentrating on Adaptive Cruise Control (ACC) and collision warning, with features such as 'collision avoidance' or 'autonomous driving' still on the list of things to come. The main aim of an electrically scanned millimeter - wave automotive radar is to develop a stable sensor that detects the range, relative speed and angular position of obstacles near and around the vehicle in any weather conditions. The radar system achieves it goals using the follow principle:

• Detection of range: utilizes the time lag of reflected waves • Detection of relative speed: utilizes the frequency shift of reflected waves • Detection of angular position: utilizes the phase differences between signals received

by multiple antennas Utilizes millimeter-waves, the transmission of which is affected less during rainy conditions as compared with laser beam (electromagnetic wave with free space wavelength of 1-10 mm) The main functions of automotive radars are:

• Navigation - the radar indicates road bends and intersections, even in bad weather. Coupled with an electronic road map and perhaps a GPS receiver, it is also able to give directions to your destination.

• Collision warning - the radar continuously scans the area ahead of the car and takes appropriate action when a collision is imminent. A collision may occur with stationary

20

objects found on the road, with cars driving in the opposite direction, or with a car driving in the same direction but slowing down. 'Appropriate' action can be the activation of an alarm or even the hitting of the brakes. However, implementation of the latter function will depend on local legislation. Usually, all responsibility rests with the driver of the vehicle but, if such a drastic interaction is done without the driver's consent, they may plead 'not guilty' and claim lawsuits against the car or radar manufacturer.

• Cruise control - the radar maintains either a preset and constant speed, or a constant distance to the car ahead.

• Airbag pre-crash sensing - if it is determined that a collision is imminent, then the radar may set off the airbag with an optimized timing, even before the hostile object has made contact with the car body.

• 'Stop and go' functionality - this is an adaptation of cruise control for lower speeds in dense urban traffic. The driver may relax and read a newspaper while an observer would believe the car was being towed by the car in front.

• Parking aid - a minimum clearance between the boundaries of the car and any other object can be maintained because the radar starts buzzing and blinking well before contact is made. Depending on the implementation, even the car's sides can be put under surveillance.

2.3.1 Requirements There are two sets of requirements for automotive radar. Cruise control and other 'highway' functions demand the capability to measure unambiguous ranges of around 200m and velocities from zero up to some 500km/h if a scenario of two Porsches shortly before collision were set as the worst case. The radar should cover some 8° to 15° either side of the front direction, to be aware of adjacent lanes and to follow road bends. A phased array antenna would be desirable because mechanically-scanned antennae are subject to wear and tear. However, phased array antennae operating in the frequency band 76 - 77GHz are still quite expensive. Angular resolution is a critical parameter and some 0.1° are deemed necessary.

The second set of requirements applies to traffic in urban environments. Velocities from zero to 140km/h can be expected. Angular resolution is less critical but the radar must cover significantly more than some 15° of azimuth when driving in a stop-and-go situation. Full 360° azimuth coverage is required when the radar is supposed to sense a side-crash too. The maximum range is of the order of 50m, but distances of a few centimetres must be measurable when the radar is used as a parking aid.

2.3.2 Problems Big and Small Targets Lorries are big targets, motorbikes and bicycles are small ones. The radar must be able to see them both, especially when an approaching lorry is being overtaken by a motorbike and the echo of the latter appears before the big return originating from the lorry. An even more complex task for the signal processor is when the same situation happens within a tunnel where the concrete walls are the most prominent source of reflection. Detection of Obstacles At first glance, anything that features different reflection properties than the concrete of the road must be either another vehicle or something dangerous. Examples are sheets of ice, rabbits, kangaroos or goods that have fallen off some lorry. However, this approach is too simple. Road surfaces can exhibit lots of non-regular features

21

that radar can easily detect, but do not represent danger. Such non-obstacles are, among others: road markings (especially within construction areas), changes of the material the road surface is made of, duct covers, and metal bars that protect the air gap where a bridge begins or ends. A device that announces every such anomaly as an obstacle would not be successful on the market because of its high false alarm rate. Mirages Layers of hot air and water puddles reflect radar signals just like they reflect light. Hence what the radar 'sees' is a upside-down mirror image of the scene further ahead, rather than the puddle itself. Therefore a puddle can be easily mistaken for a very deep hole in the road. On the other hand, the signal processing may take advantage of mirages because they provide a means to 'look around' the car in front by examining the road under its belly from a grazing angle. Bridges and Tunnels Imagine a scene where the road runs over the top of a hill and there's a bridge that spans the road at right angles. At first glance, the bridge would appear like a wall that has been placed across the road. The free space under the bridge (or the hole that makes up the tunnel) becomes visible only at closer distance. There isn't much time for the radar to take a decision if you consider speeds in excess of 250km/h, which aren't uncommon on Autobahns in Germany. Without second source information this situation will lead to a false alarm. The second source could be a digital road map stored in the vehicle's navigation system with extra records of these kinds of features. Obviously, one specific feature is lacking from all automotive radars currently in development: no manufacturer has included the capability to warn against police radar. Thus, the avoidance of speeding tickets is still left to the driver's responsibility. No matter how far radar can see through fog and haze, it doesn't influence the quality of a car's brakes or the friction between tyre and road. Like any other part of a car, the radar can break at any time without warning, and no manufacturer is able to guarantee that their equipment has 100% detection capability. Thus, you should take the information derived from sradar as a means to increase safety, not speed. It is interesting to note that bats' acoustic sensors have long been employing continuous wave and chirped pulses. Moreover, bats have also developed suitable means to avoid mutual jamming when hundreds of them are simultaneously active in a given place. Radar manufacturers still have to think about that.

2.3.3 Analysis of Systems In the follow sections will analyze automotive radar systems that are to be found in the passengers vehicles. The aim is to give the read good knowledge about how automotive radar systems work and how they are implemented.

2.3.4 Active Cruise Control Active Cruise Control, or Adaptive Cruise Control, takes the normal cruise control to another level. This system not only maintains the desired speed chosen by driver, but also monitors and controls the distance to the vehicle ahead of the car on the motorway or a country road. As long as the lane ahead of the car is free, ACC works pretty much in the same way as a conventional cruise control. But as soon as another vehicle ahead is within a certain distance and driving at a lower speed, long range radar, mounted in the front, detects the situation and ACC adjusts the distance by braking the car the exact amount that’s needed (see fig.26&27).

22

When activated, ACC give gas and slightly applies the brake in a way to keep as high comfort as possible. The driver can however take over the control at any time or apply the brakes himself. And this is indeed necessary whenever the system reaches its limits, since it is for most car manufacturer’s philosophy to maintain the driver’s responsibility in and for his car. To activate ACC the driver first chooses his "personal" speed in 10 km/h intervals (see fig.28).

Fig.28 When approaching a slower car, the preprogrammed distance is shown in the speedometer. ACC is created for use between 30 and 180 km/h, the speed chosen by the driver being marked on the speedometer and constantly maintained as long as the vehicle ahead does not require a reduction in speed. The technology behind the system is a 77 GHz long range radar sensor, serving as the key component in the system. The radar is able to detect vehicles ahead at a distance of up to 120 meters, largely independent of weather conditions (see fig.29).

Fig.29 The radar is mounted under the front bumper. As soon as the driver’s car approaches another vehicle from behind, ACC adjusts the driver’s speed smoothly and precisely to the car ahead, at the same time keeping a constant distance the driver is once again able to choose himself from four levels on the control lever. If the slower car ahead exits the road, ACC automatically increases the speed until the car is driving in the same speed as it did before the slower car showed up.

Fig.26 vehicle with ACC approaches car in front. Vehicle speed is set to 70 mph; ACC detects slower cars in front and adjust speed automatically

Fig.27 ACC takes action, automatically adjusting the ACC vehicle's speed to

match the target vehicle's speed. If the ACC vehicle loses its target (for example

during a lane change) then the ACC vehicle will automatically reaccelerate

to its 70 mph set speed.

23

The amount of power applied on the brakes is limited to a comfortable level that corresponds to a car deceleration of 2 m/sec2. This amount is quite sufficient for fine adjustment of the car’s speed and distance in maintaining the system function. But if a situation occurs that needs the system to apply the brakes the maximum amount, for example if the car ahead panic brakes, the system can not take this decision. That might result in the car brakes on its own even if the driver has other intention and an accident occurs by a result of the over taking system. Instead, the driver is informed accordingly by a light and sound. ACC relieves the driver of the permanent and monotonous and unpleasant task of constantly adjusting the car’s distance and speed. This is a particularly convenient when driving on motorways or expressways with road speeds constantly changing. In these situations, conventional cruise control is often useless. Features

• Assists with driver overload and stress in congested highway situations. • Assists with key driver functions (accelerating and braking) while cruising. • Maintains a set distance automatically between vehicles. • Serves as a cornerstone to advanced safety systems.

Specifications

• Dimensions 95mm x 95mm x 63mm Distance measurement:

• range 1m ... 200m • accuracy ± 5% or 1m

Speed measurement • range ± 250kph • accuracy ± 0.1kph

Lateral position measurement • range ± 6° • accuracy ± 0.3°

2.3.5 Adaptive Cruise Control with Stop & Go With additional Short Range Sensors (< 50m) the speed range can be extended to 0kph if they cover the whole width of the lane in front of the car. As a result functions like Follow Stop and Stop & Go can be provided. Follow Stop & Go means that in traffic jam situations the car follows the impeding vehicle until this comes to a standstill. To drive off afterwards the driver has to take action (e. g. press a button). Stop & Go means that also drive off is done automatically. For safety reasons additional sensors are required. With the addition of vision systems (video camera) detection of lane marking is possible. .

Fig.30 the ACC vehicle approaches a target vehicle and

begins to react to it.

Fig.31 traffic jam assistant follows the target vehicle to a complete stop. It then allows the ACC vehicle, upon driver command, to follow the target at a comfortable distance in

traffic jam situations.

24

Features

• Enhancement of standard ACC by enabling braking to a complete stop • Start and acceleration initiated by driver • Short range object detection with additional sensors can allow automatic restart

without driver intervention

2.3.6 Collision Warning Collision Warning Systems can assist drivers by helping to prevent or mitigate accidents. Combining long and short range radars with a video camera, collision warning monitors the road ahead. If the car approaches an obstacle (stationary or moving) and the driver does not react, a warning light activates and is reflected in the windscreen. At the same time, an audible buzzer sounds and a brake function is automatically activated to build up higher braking pressure. In certain situations, this is sufficient to catch the driver’s attention and avoid the hazard. Some cars also tightens the seat belts, adjusts seat positions including rear seats (if installed) and can also close any open windows and the sunroof if necessary.

2.3.7 Collision Warning and Preparation With short- and long-range radar systems on board, adaptive cruise control system evolves from a convenience system to a safety system which can identify and warn the driver of an impending accident by producing an audible warning or adapted one such as a programmed vibration of the steering wheel or seat cushion. In addition to warning the driver to take action, the brake system can be readied to provide maximum brake boost once the driver does engage the brakes enabling reduced stopping distances. When the driver brakes, the system monitors the pedal pressure. If the pressure is too light for the car to be able to stop in time, the system steps in and amplifies more braking power. If the speed is not too high, this brake boosting function can help avoid a collision. In addition, the reversible electric motor in active control retractor seat belts can be signaled to remove seatbelt slack and better position occupants for a potential crash.

2.3.8 Pre-Crash Emergency Braking with Millimeter Wave Radar and Stereo Camera Fusion System Pre - Crash Emergency Braking technology combines forward looking radar and video systems to provide a very complete, accurate and real-time picture of the road ahead. Coupled with electronic stability control, these systems can be extended beyond more warning systems to automatic emergency braking. If sensors detect a potential impact a certain amount of brake pressure is applied automatically and the brake system is readied to provide maximum brake boost immediately once the driver does engage the brakes. The reduced stopping distance helps to mitigate the effects of the potential accident. The active control retractor can also be triggered prior to the actual impact thus improving the margin of safety for the occupants.

Fig.32 vehicle cruises along the highway, collision warning system

monitors vehicles in the lanes ahead.

Fig.33 collision warning system in the host vehicle senses a rapid

deceleration of the vehicle in front and alerts the driver. The host

vehicle achieves full brake force immediately upon request of the

driver through brake pre-fill.

25

2.3.9 Rear Pre Crash Safety System A millimeter-wave radar device in the rear bumper detects a vehicle approaching from behind. If the system determines a high possibility of collision, the hazard lights flash to warn the driver of the rear vehicle. And if the system determines a further increase in the possibility of a collision, it automatically activates the front-seat Pre-Crash Intelligent Headrests, which shift to appropriate positions prior to impact to reduce the risk of whiplash injury.

2.3.10 Status Of Radar Based Automotive Applications Automotive radar is a key technology especially due to its advantages like weather independence and direct processing of range and velocity when compared to alternative sensors like video, laser, and ultrasonic. Additionally radar offers the vehicle manufacturers the stylistic advantage of mounting behind a plastic bumper.

2.3.11 76-77 GHz Long Range Radar Systems ACC systems are used to relieve the driver of part of his task of keeping distance and warn him in critical situations, thus making driving is less stressed, especially in flowing traffic. The driver remains fit for a significantly longer period of time. The responsibility for the safe handling of the vehicle remains with the driver under all circumstances ant at any time. The system provides information about traffic situations ahead of the vehicle, thus making it possible to react to altered traffic conditions by accelerating, braking or changing gear. The driver is given visual information and in case of a critical situation an acoustic warning. Typical situations in which ACC can be used are on motorways or dual lane ways, where the driver has a lot of load taken from his shoulders, especially when holding station within the traffic flow. ACC can be activated typically at speeds of 30 km/h to 180 or 200 km/h. Typically ACC systems are mounted in the radiator-grille or front bumper, operating in the 77 GHz band. ACC systems from ContiTemic A.D.C. are available in different Mercedes models and in the ACTROS truck. But also other car manufacturers and suppliers like Jaguar or Cadillac with Delphi, Volkswagen with TRW, BMW or Audi with Bosch or Volvo and MAN in their trucks with TRW offer ACC for highway operation as an option and others plan to do this in the near future. Based on a 76 GHz LRR sensor Toyota offers even a precrash system in the Lexus RX 330 from Denso working as a brake assist and reversible belt pre-tensioner since 2003. Honda developed a 'Collision Mitigation brake System' (CMS) based on Fujitsu Ten’s 76 GHz radar for predicting rear-end collisions and controlling brake operations in the Inspire.

Fig.34 adaptive cruise control and collision

warning systems monitor the road ahead

Fig.35 adaptive cruise control and collision warning systems sense an impending vehicle

sending signals to the vehicle's braking system to prepare for emergency braking

Fig.36 emergency braking begins. The brakes apply full

pressure

26

2.3.12 24 GHz Short Range Radar Short range radar sensors can enable a variety of applications as depicted in Fig. 14:

• ACC support with Stop&Go functionality • Collision warning • Collision mitigation • Blind spot monitoring • Parking aid (forward and reverse) • Lane change assistant • Rear crash collision warning

Especially the combination of LRR and SRR provides valuable data for advanced driver assistance systems. In situations where drivers are forced to brake, the Brake Assist Plus system will calculate and generate the braking force needed for a given situation within fractions of a second based on the radar information. Here two radar systems are combined to monitor the traffic situation in front of the vehicle: newly developed ultra-wide band short range radar based on 24 GHz technology works together with the proven 77 GHz cruise control system [17]. Whereas the LRR radar is designed to be able to track three motorway lanes over a distance of up to 150 meters with an angle of 9°, the new SRR uses an angle of 80° to monitor the immediate area up to 30 meters in front of the vehicle (Fig. 15). Preventing rear-end collisions against the preceding car is thus fulfilled. The SRR sensors are based on a pulsed radar concept. The front end consists of the transmit circuitry, the receive circuitry and the control and processing circuits. An object is detected by measuring the elapsed time between a transmitter pulse and a correlated received signal. With this correlation receiver architecture a detection range of 0.2 to 30 m, a range (object) resolution of 15 cm and a range accuracy of 7.5 cm can be achieved. Up to 10 objects with range, bearing and velocity information can be classified [18]. The individual sensors are connected via a local network to the radar decision unit, which is on its part connected via the car controller area network (CAN) bus to the different electronic control units of the car. SRR is also developed by other suppliers like TDK or Siemens VDO.

2.3.13 Future Radar Developments 77 GHz LRR Systems The 76 – 77 GHz band is recognised widely by overseas regulatory bodies, and by international and regional standards bodies, for automotive radar applications. The interference risk presented by automotive radar applications to other potential users appears minimal, noting that the radioastronomy and space research communities may need to take some precautionary action locally at some future time. Nearly all well-known automotive radar suppliers like ContiTemic A.D.C., Bosch, TRW, Delphi, Fujitsu Ten, Mitsubishi Electric, Denso, or Hitachi are working on sensors for next generation ACC systems. Thereby further development activities are focused on reducing sensor and sensor adjustment cost on one hand. On the other the aspired goals for the sensors are a higher distance range from less than 1m to up to 200 m, up to ±10° opening angle in

Fig.38 combination of long range radar and short range radar for advanced

safety features

Fig.37 possible applications using short range radar

27

long range and a relative velocity range of up to ±260 km/h. Hidden cars covered by preceding vehicles are detected by ground clutter reflection based on higher sensitivity. The ACC-Sensor can also be run in an narrowband pulse mode. This mode has not the high detection range and sensitivity of the narrowband FMSK-mode, but is advantageous in classical STOP&Go situations, with many targets with low relative velocity in near range. If such a situation is ahead, the sensor will automatically switch in the pulse mode with a higher range resolution. If such a dense traffic situation is resolved and the eco velocity goes up, the sensor changes back into the FMSK mode with its high detection range. The possibility to switch the mode makes it possible to use the ACC sensor even in dense urban traffic situation sand the minimum activation velocity for ACC-systems can likely be reduced to zero. It was shown that it is feasible to produce an ACC sensor operating at 24GHz. It certainly does not have the full performance of its 77GHz counterpart, but the 24GHz ACC function is robust enough in practice on test cars. The potential of that new sensor is high, as it allows a significant cost reduction. 79 GHz SRR System and Technology Specifications for a 79 GHz SRR sensor are as follows:

• Frequency 79 GHz • Bandwidth 4000 MHz • Maximum field of view ± 80° • Range 30 m • Range Accuracy ± 5 cm • Bearing accuracy ± 5°

With higher frequencies semiconductor power output decreases, parasitic effects (steel losses, Joule losses, copper losses) are more stringent, and packaging and testing are more difficult. The transition from 24 GHz to 79 GHz causes an increase in frequency and a reduction of wavelength by a factor 3.3. The smaller wavelength enables reduced antenna size and spacing (~λ) and lower effective antenna area (~λ2). The higher frequency yields increased atmospheric and bumper losses. The main challenges for 79 GHz SRR technology are given below:

• low chip and component costs • low assembly costs • improved performance • reduced power consumption • improved electrostatic discharge (ESD) and electromagnetic interference (EMI)

properties • high cycle times / update rates

2.4 Automotive Laser Scanner Laser sensors increase safety in automotive traffic. They are able to capture, with extreme accuracy, data in the vicinity of the vehicle and can thus influence the driver and vehicle. In contrast to other technologies, laser scanners have a very high area of visibility at a high degree of angular resolution and accurate measurement. The laser scanner sensor enabling multiple applications through a single device: Automatic Emergency Braking, ACC Stop&Go, PreCrash, Traffic Jam Assistant, Pedestrian Safety and Lane Change Assistant etc. Even in very poor weather conditions, laser scanners function with a high degree of precision and reliability. In addition, they can be manufactured cost-effectively. This makes it possible

28

to produce a “safe vehicle” in large quantities, bringing the vision of accident-free driving closer to reality. On the basis of a fully rotating infrared beam, ALASCA allows the scanning of a large angle of the vehicle environment. The detection range of 30 m (5% target reflectivity) is ideal for all applications requiring a high-precision angle and distance resolution in the short and medium range.

System Characteristics

• Infrared measuring system • Detection of multiple targets • Compensation of pitch angles • High-accuracy distance measurement • Detection of the object contour

Specifications

• Range 0.3 m to 80 m (30 m on 5% reflecting target) • Range resolution ± 5 cm • Rotational frequency 10 ... 40 Hz • Vertical angle range 3.2° subdivided into 4 levels • Horizontal angle range Up to 240° (depending on the mounting position) • Horizontal angle resolution 0.25° ... 1°

Principle of Operation

• A infrared diode generates a short light pulse • A rotating mirror transmits the infrared beam • The target reflects the infrared beam • The photodiode receives the reflected beam • Time-of-flight measurement supplies the object distance • The angle resolution is supplied by the angle-encoder of the mirror drive • Calculation of object velocity and acceleration • Object tracking depending on the desired application

Fig.39 Automotive laser scanner

Fig.40 Signal processing

29

2.4.1 Signal processing As an illustration, the scan data are shown in the above driving scene, the four coloured dotted lines representing the scanning levels. These four levels are always in use to avoid object losses caused by the pitching motion of the vehicle under various load, acceleration and deceleration conditions (typical +/- 1.6°).

ALASCA calculates the object outline from the scan data by appropriate algorithms. The contour lines serve to extract and classify the objects. In addition, the distance, direction and relative velocity data of each object are supplied for further signal processing. For security reasons the driver is able to overdrive the automatic control at ACC Stop&GO. The detection ranges depend on the sensor mounting position. With one sensor in a central mounting position, an angle of 120° can be covered, e.g. for the presented applications Stop & Go and Collision Mitigation. With two sensors mounted at the left and right side of the bumper, and multiple scanning of the ROI (Region of Interest) in front of the vehicle an enlarged angle of 240° can be covered. The advantage is the additional monitoring of the side areas of the vehicle. Therefore, miss-operation due to reflections caused by fog or the road surface can be avoided at close ranges of less than 20-30 m. The reasons for the decline seen in detection performance include broken or dirty reflectors on preceding vehicles and reduction of the optical power of the received light pulse due to exhaust emissions. The scattering or absorption of light by rain or fog can reduce the detection performance by around 30 % compared with the performance obtained in fine weather. Two types of beam transmission systems are currently used for laser radar. One is a fixed beam system, in which the orientation of the transmitted beam is constantly fixed. The second type is a scanning system, in which a narrow laser beam sweeps a pattern of a certain specified angle in the direction of potential objects. There is also a multi-beam variation of the fixed beam system, which employs several fixed laser beams in order to expand the scope of detection (Figs. 20 and 21).

Fig.41 vehicle detection performance

Fig.42 factor degrading detection performance in rainy weather (change

in maximum detection range)

Fig.43 Multi – beam system Fig.44 Scanning system

30

One advantage of the scanning system is that it can provide satisfactory detection on curves. Another advantage is that it gives more detailed information than a multi-beam system on the relative positions of a radar laser equipped vehicle and a detected object. This is possible because, in addition to detecting the distance to the object, it also detects the angle between the vehicle and the object. A lot of work has already been done to detect the lane and road borders. Several approaches use vision based sensors, especially CCD-cameras (e. g. [19] , [20]). In case of daylight, good weather conditions and white border lines, this approach reaches reasonable results. But in non cooperative areas and during night or in bad weather conditions, the white lines are not always visible, so this approach fails. A laser scanner is nearly independent on the hour of day (night or day) and it is a high resolution measuring system: it is able to detect white lines as well as deliver detections on roads without boundaries (guardrails, etc.). Typical reflection values are shown on the left side of Figure 22.

Using the range values for detection, the street appears as part of a straight line. On the right side of figure 22 a typical scan, transformed into a Cartesian coordinate system, is shown. So the detection algorithm can be divided in two parts: the first part estimates one or more possible lines, the second part defines the segments. The output function of the signal processing compares the road data with estimated tracks holded in the ECU, then there is a matching process; the outcome is in the Figure 22. Depending on the sensitivity of the laser scanner, the road border can be detected at further distances and in low reflecting situations.

2.5 Ultrasonic Sensors Ultrasonic sensors are effective under the bad weather condition where the optical sensors are ineffective. However, they have a fatal weak point that their longitudinal detectable range is very short. Sometimes, walls located at both side of ultrasonic sensor cause a miss-detection of objects. Their typical environment-sensors are laser range finder, CCD camera, millimeter wave radar, ultra-sonic sensor, etc. [21]. Though optical environment-sensors such as laser range finder and CCD camera are very effective under fine weather condition, they cannot use them under bad weather condition, in particular, in a dense fog. Millimeter wave radars are effective under such bad condition, but it is expensive. Thus a low cost environment-sensor that uses ultrasonic sound wave. Ultrasonic environment-sensors have very simple structure and are effective when cannot use the optical sensors. They can detect objects even if the fog is very dense. Because their longitudinal resolution is high enough, we can measure a distance between sensor and object with high resolution. Although ultrasonic sensors are popular environment sensors in collision warning systems [22], their lateral resolution, in other word, the directional resolution is very low. The direction and distance of objects are detected from

Fig.45 Laser tracking

31

the phase difference among reflected waves that arrive at several receivers. The most important weak point of the ultrasonic environment-sensors is the shortness of detectable range in longitudinal direction. If walls are located at one side or both sides of the ultrasonic sensor, reflected waves from objects and walls arrive at one receiver simultaneously. This reflected waves interfere with each other and cause a miss-detection of objects. Because side lobes of steering characteristics cause miss-detection of objects, it is important to decrease the height of side lobes.

2.6 Example of Fully integrated cars Here follows some examples of cars that have been fully integrated with different remote sensing equipment

2.6.1 The Springrobot The hardware systems installed on Springrobot platform consists of brake-by-wire and steering-by-wire electronic actuators, sensing and fusing computer, control computer, vehicle ego state sensors, and road environment sensors. The vehicle control systems act as the interface between the control computer and the body of the vehicle. One is used to control the steering wheel, and the other is used to control the throttle and brake. The vehicle could switch very naturally from manual operation to autonomous control in second, even in an emergency. It is well-suited for applications in vehicle warning and control systems, such as forward collision avoidance assistance systems, adaptive cruise control systems, lane departure warning systems, etc. The Springrobot intelligent vehicle has several different types of environment sensors including a

• EATON EVT-300 millimeter wave radar, • SICK LMS221 lidar, • Two TMC-9700 colour progressive cameras, • Dual-axis acceleration gyro, wheel encoder, • NOVTEL DGPS(Proak-G2-Ll/L2,RT2).

A SCIK LMS221 laser measurement system is installed on the vehicle that provides a 180-degree scanning field, with a resolution of 0.5 degrees, a maximum range of 80 meters. Of course, the actual distance of detected object depends on the reflectivity of the object, and it could detect not only still objects but also moving ones. For vehicle active safety systems, it needs farther range measurement sensors, then EATON EVT-300 millimeter wave radar is installed in the center of the vehicle's front bumper. It provides a 12-degree scanning field and a average range of about 350 feet, and detects only moving targets relative to then subject vehicle because of the use of Doppler technology. The DGPS consists of the base station and mobile station, and the base station comprises the ProPak-G2-LI/L2 double-frequency receiver,GPS-700-L/L2S double -frequency antenna, and PDL broadcast station. Similarly, the mobile station comprises the ProPak-G2-RT2 double frequency receiver,GPS-700-L/L2S double frequency antenna, and PDL broadcast station. Many errors of GPS receiver are common in a local area, then DGPS utilizes this characteristic and attempts to compensate by using a base station situated at a precisely surveyed position. This DGPS can provide the position resolution with accuracies up to 2cm and update rates up to 20Hz.Moreover, stereovision system utilizes two PULINX TMC-9700.

32

Fig.46 Hardware and software structure

2.6.2 Mitsubishi Active Safety ASV The ASV is equipped with so-called "Multi-Eye System" which consists of two lane-detecting cameras, two scanning laser radars, six passive trigonometric-type sensors, and three stereo cameras, to detect surrounding traffic situations. The "Friendly Man-Machine Interface" consists of sophisticated indicators, warning systems, and indirect visual field systems using small video cameras. The collision avoidance system assesses the situational risk and warns the driver or performs an automatic collision avoidance if necessary. Scanning Laser Radar Two scanning laser radar are installed in each side of the front bumper. Their positioning helps to detect cut-in vehicles earlier. They cover the wide front area almost in duplicate for redundancy. The scanning mechanism is composed of cam-driven mirrors and a stepping motor. It provides not only long measurable range, but also precise angular positioning.

Fig.47 Scanning Laser Radar Lane-Detecting Camera Two small video cameras are installed behind the rear view mirror to detect both side white or yellow lane markings of the current cruising lane. A wide 1/3" CCD camera is employed for the short range, and a normal angle camera for the long range. Lane marks are detected using image data processing techniques, including edge extraction and processing area restriction based on some a priori knowledge of lane mark positioning.

33

Passive Trigonometric-Type Sensor Six passive trigonometric-type sensors are installed in the body panels around the ASV. The detective part of the sensor is a pair of photo-diode arrays arranged vertically. An auxiliary illuminating LED with random stripes is combined with the sensor, and is automatically activated when the sensor views a low contrast and/or dark target. Side-Rear 1 Rear Stereo Camera A side-rear stereo camera is installed in each front fender, and a rearview stereo camera beside the rear license plate. The ranging principle is based on trigonometry, using the vertically-arranged stereo vision. Nine windows are horizontally arranged in the camera view, and ranging calculations are executed only within the window areas. The side-rear stereo camera has the unique stair-shaped window arrangement which effectively detects other vehicles in the adjacent lane. The total amount of the window area is only13% of the whole camera-view.

Fig.48 Friendly Man-Machine Interface The CO-Driver System uses an easy-to-use and versatile "Friendly Man-Machine Interface" to communicate with the driver. The Head-Level Multi-Function Meter is placed at the lower edge of the windshield where the driver needs so little line-of-sight movement to see it as a head-up display, and it does not deteriorate his forward vision. The meter displays vehicle speed under normal conditions, a three-layer operation menu when the driver touches the control stick, and warnings when a hazard occurs. A Multi-Communication Display is placed in front of the driver in the instrument panel, which shows more precise information, such as a navigation map and surrounding obstacle status. Blind-spot vision is provided for the driver by small video cameras.

Fig.49

34

Actuators for Automatic Collision Avoidance When the CO-Driver System detects potential danger, it sounds a warning to alert the driver. If the driver fails to respond appropriately or quickly enough, however, the CO-Driver System activates the automatic brakes and/or automatic steering to avoid a collision.

Fig.50

2.7 Integration of active safety and remote sensing Many of the remote sensing systems are based on existing active safety system. The remote sensing system handles a dangerous situation as a perfect driver would handle it. But since nobody can reacts as fast and accurate as sensors and actuators these remote sensing system are necessary. The ABS-system can be used in the Pre-Crash Emergency Braking. In a dangerous situation the car can stop completely and to make this as fast as possible, the ABS is necessary. The Adaptive Cruise control is based on an ordinary cruise control but the radar system adds an amazing new feature. Lane keeping aid and Parking assistance (a system that parks your car for you, very useful when parallel parking) uses power steering when the car it self can turn the steering wheel. The power steering can also be used in collision avoidance then the sensing systems become accurate enough

35

3. Competing technologies We use this term telematics to describe the process of long distance transmission and computer based information. The principle is to exchange safety information thanks to an electronic sub-system in a car. These informations can be about road hazards, location and speed of vehicles. The vehicle-controlled collision avoidance can be categorized into independently controlled, cooperatively controlled, and coordinately controlled vehicle groups. This table explains the differences.[27] Control mode Control and

Decision Making

Controller Accessibility

Information Exchange

Communication mode

Independently Controlled

Make own decision

Not accessible by external controller

No exchange of information

No inter-vehicle communication

Cooperatively Controlled

Use common strategies

May be partially accessible by external controller

Allows limited information sharing

One way relay mode, or bidirectional inter-vehicle communication

Coordinately Controlled

Master-slave type. The master vehicle controls all

Accessible by the external controller

Allows extended information sharing

Bidirectional inter-vehicle communication

Vehicle telematics can be used for car accident prevention, particularly with short range communication in vehicle early warning. There are really many potential applications for vehicle telematics. Telematics refers to a lot of different systems, for instance emergency warning system for vehicle, wireless safety communication and automatic driving assistance systems.[28] Short range radio links are used, so temporary ad hoc wireless local area networks has to be created. To achieve these requirements, vehicles and fixed locations like traffic signals and emergency call boxes will receive wireless units. The information is provided by sensors in the car, at the fixed station and maybe connections to wider networks. Even without fixed units, news about fixed hazards can be maintained by moving vehicle by passing it backwards. The traffic lights can be smarter as well, by using the informations to reduce the chance of collisions. All these informations have to be displayed to the driver in some way. In a distant future, there will probably be some links with cruise control or other vehicle control aids. Then, cars behind a column member who slow down will automatically slow down as well. Telematics technologies can also provide warning informations to surrounding vehicle in the vicinity of travel, intra-vehicle and infrastructure. So, instantaneous direction travel cognizance of a vehicle may be transmitted in real time to surrounding vehicles traveling in the local area of vehicle equipped to receive warning signal of danger. For wireless wide-area networks there are mainly two available technologies: data transmission over cellular networks, whether analogue or digital, and data transmission over mobile data networks. The main difference between these technologies is the data transport

36

mode. Cellular networks generally utilizes circuit switching technology, whereas mobile data networks employ packet switching technology. The many major wireless voice carriers have plans to move to cellular networks over next one to three years. Currently, due to physical layer constraints, wide-area networks typically feature low-speed wireless data transmission, on the order of 9.6 Kbps. However, with the emerging new protocols, much higher data transmission speed is supported. [29] Here is Toyota Key technologies for its Intelligent Transport System (ITS). The goal for the ITS program is to obtain a completely autonomous car.

Fig.51 Intelligent Transport Systems Key Technologies [30]

AHS (Advanced CruiseAssist Highway System) is a system intended to reduce traffic accidents and congestion through vehicle-highway coordination using sensors, vehicle-to-infrastructure communication, and other advanced ITS technologies.

Fig.52 AHS principle [31]

37

Fig.53 Future Image of ITS Information Communications System [32]

The advantages of inter-vehicle communication system are to alleviate traffic congestion, to ease drivers' stress, to reduce traffic accidents at intersections and to lessen driver's burden of making decisions. The main disadvantage of these systems is that all cars and the surrounding environment need to be equipped. The consequence is that it will probably take a long time to introduce these telematics technologies, since the positive effect can only be seen when the whole car park and driving environment will be ready.

4. HMI The remote sensing systems discussed earlier and the systems they interact with or are a part of have the purpose to increase safety. They can either inform the driver of hazards or intervene through braking or other actions. In both cases this will have an effect on the driver. HMI (Human Machine Interface) are theories that in automotive applications describes how the driver reacts to information from the car and what he does with the information.

4.1 Autonomous and informing systems Autonomous systems that take control of the car and prevent accidents or work as an autopilot is today not in a near future. Even if the required sensors and systems can be developed and manufactured the price would be high. This is though not the main reason for not doing it, but if a car would be able to make decisions it can also be sued. This means that if something goes wrong the car manufacturer is responsible. As long as the systems only recommend the driver to do something in a situation it is the driver that’s responsible. There isn’t only the problem then the system don’t recognizes a hazard and the driver that already rely on the system don’t react, also the system can be false activated and take control over the car making a situation out of nothing. To prevent this type of malfunction tow sensing systems can be

38

used, but it also makes the system more expensive. Therefore it will take time before we se this type of autonomous active safety in cars.

4.2 Warning signals and information systems The driver can react on signals from an active safety system in different ways. Depending on the type of signal and how susceptible the driver is to the signal he can apprehend it slower or faster. The reaction time is crucial for the outcome of the situation. If a signal is given too late it won’t help the driver. Instead if it is given early some drivers will be pleased with deciding if it is reasonable and correct or not, other drivers will however find the early signal annoying, thinking that it newer became a situation and therefore shutting the system down. Until today most signals given to the driver have been light and sound signals. New systems like GPS require new types of information with signs and pictures on screens. Information on screens often requires focus of attention to be moved of the road. Speech synthesizer can help the driver to keep looking at the road, but comes often only English-speaking and is considered annoying by some drivers. The head up display (HUD) has been used in combat airplanes and is now on advance as an interface also in cars. It projects the picture with information directly on the windscreen, so that the driver doesn’t have to move his eyes. In some cars the driver can choose what information to be displayed depending on road type and traffic conditions. The HUD can be one of the better ways to inform the driver of hazards in a future integrated active safety system. Some applications will be discussed later in this report. The reaction time from signal to action is not only apprehending the signal but also realizing what’s best to do. This time is strongly dependent on how used the driver is to the situation and how expected it was (Mikael Ljung). Some reactions don’t have to be managed consciously but acquired by experience and therefore have a very short reaction time. With this in mind it is likely to say that systems of remote sensing like forward collision warning system in Volvos is working because drivers are used to brake for things in front of the vehicle. To be able to develop a active safety system that interacts well with the driver it is important to use the right analyze methods. For example in traffic testing or driving simulators can be used. Also human behavior models can be used to better design active safety systems. The latter is described further in appendix A.

4.3 Driver performance and user friendliness Like computers, the human has a working memory (Baddeley & Hitch) which limits the number of simultaneous tasks that can be performed. The tasks also compete for the attention from the driver. This is a typical cause to accidents when we talk about in car phones or in car entertainment such as music players. However it can be a cause to problems even if the system that brings the information was originally meant to prevent accidents. For example if front collision warning tells you that the car in front is to close, and you actually where getting that close to make a maneuver you get disordered and do something unexpected. A possible problem in the future may be with camera assisted night vision that informs the driver of what is coming up on the road. If the driver concentrates too much on the display with night vision, he might miss things on the road not displayed. Solutions to this problem and other problems will be discussed in another part of this report. Still it’s more common that these problems appear while using other, not driver related systems. It has been shown that accidents happen then the driver performance doesn’t fulfill the demands from the

39

environment (Shiner). This is likely to happen then a sudden increase in performance level follows a lower level as shown in fig.54 point B and C

Fig.54

It is likely in a situation as in point B that the driver tends to do other not driver related things like changing station on the radio. In situations like that a lain departure warning system or other active safety systems can be very helpful. These systems alert the driver and take his attention back to driving. However they have other HMI related disadvantages that will be discussed further. After a while then the driver gets used to the warning system he tends to use the safety margin for other purposes. If the car is equipped with for example lain departure warning, the driver takes more time doing other things, not looking at the road, because he knows that he will be warned if he passes a line. The problem appears then the system misses or there isn’t a line to detect and the car goes into the ditch. Other problems that come with some systems and are more unexpected are wrong usage of them. For example go in reverse without looking back because you have parking sensor. Or not looking over one’s shoulder if the car is equipped with blind spot detection. The car manufacturers can disclaim there responsibility in some situations if they can predict them. However this misusage can be dangerous and that’s not what the manufacturer intended. It is also required, for the active safety system to work, that the driver doesn’t shut it off. To prevent this, the interface must be good enough and the system itself must be reliable. If the system misses a risk situation or warns then where isn’t a risk the driver won’t buy the system. The error frequency can’t be too high but also a few errors can be dangerous if the driver has started to depend on the system as discussed earlier.

5. Analysis of traffic incident types There is no doubt about traffic being a dangerous undertaking. Statistics again and again shows huge numbers of people getting seriously injured or even killed by, or as a consequence, of traffic. In this section a deeper look at traffic accident statics will be conducted, trying to identify certain key types of accidents that either has a large quantity of fatalities and injured or situations that can be helped off by designing new safety systems.

5.1 What does that statistics say? Every year most traffic road administration from different countries publishes data concerning traffic and traffic accidents. The Swedish Road Administrations publication is somewhat brief and don’t give the wanted insight on types of accidents and cause. Therefore the American counterpart, NHTSA’s yearly publication, and other studies conducted on their behalf, will used as reference to get a good picture of what is going on in today’s traffic.

40

It becomes clear after reading the works from NHTSA that the most dangerous and frequently types of accidents are: Rear End collisions, Intersection collisions and Pedestrian collision. These all are areas that have a lot of potential, and work can be done to prevent and mitigate a big portion of these accidents, thus saving valuable lives. Short facts from “TRAFFIC SAFETY FACTS 2005” by NHTSA

• More than 6.1 million police-reported motor vehicle crashes occurred in the United States in 2005.

• A total of 43,443 people lost their lives in motor vehicle crashes in 2005. Another 2.7 million people were injured.

• The majority of persons killed or injured in traffic crashes were drivers (64 percent), followed by passengers (28 percent), motorcycle riders (3 percent), and pedestrians (3 percent).

• Of the persons who were killed in traffic crashes in 2005, 39 percent died in alcohol-related crashes.

• Compared with other vehicle types, utility vehicles experienced the highest rollover rates in fatal crashes, 35.4 percent

Type Fatalities* Injured* Percent of all FatalitiesRear End 2 118 513 000 5.4 % Intersection 8 655 847 000 19.9 % Pedestrian 4 520 59 000 11.5 % Table1 Traffic accident statistics *Data collected from [23, 24]

5.2 Why do accidents occur? There are many different reasons to why accidents occur, but this section will serve to give an improved insight of what the problems might be. Information is taken from “TRAFFIC SAFETY FACTS 2005” by NHTSA. Here follows a list of reasons that might cause accidents. Drowsiness

• Drowsy • Sleepy • Asleep • Fatigued

Inattentive

• Due to use of car phones • Distracted by children • Lighting a cigarette • Operating/adjusting radio • Reading, eating, talking, applying cosmetics, using electric razor, painting nails, etc.

Vision Obscured by

• Rain, snow, smoke, fog, dust • Reflected glare, bright sunlight, headlights • Curve, hill, or other design features • Building, billboards

41

• Trees, crops, and vegetation • Motor vehicle, parked vehicle • Splash or spray of passing vehicle • Inadequate defrost, defogging, lighting systems

Devices in Vehicles with Potential for Distractions

• Cellular telephone • Computer, Fax machines • Onboard navigation systems • Two-way radio, head-up display

5.3 Three accident types We will look at three accident types to be able to compare the remote sensors described earlier. The accident types are representative for several reasons.

5.3.1 Frontal collision in intersection Frontal collisions are common and deadly. They happen for different reasons on different roads. We have chosen to look into one type of frontal collision, partly because it has different reasons for happening and also it alone stands for 20% of the fatalities in USA. The car coming from beneath in the picture, making a turn, can miss the other car for more then one reason. The driver can miss judge the other cars speed and think that he will make it through the intersection. Or be looking in other direction just before the intersection and therefore not se the other vehicle. Also some conditions like bad whether or the sun standing low and towards the driver that is turning. This type of accident can be prevented with some type of remote sensing system. It has to have a long range and handle bad weather conditions which one is best suited will be discussed later.

Fig. 55

5.3.2 Proposed solution for frontal collision in intersection In an accident like this it is important that the safety system can detect the accident at an early stage, due to the high speed that is involved. The worst case scenario could be that the approaching car is driving at a speed of about 110 km/h and the turning car at a speed of about 40 km/h. This kind of intersection is most likely to be found on roads with a speed limit not higher than 90 km/h, so a worst case speed of 110 km/h would be a good assumption. In the same way, a car that is about to turn will most likely not drive faster than about 40 km/h. This gives a total speed difference of 150 km/h which is equal to 41,67 m/s. A safety margin of three seconds will result in that the safety system has to recognize the car 125 meters away. Two different safety systems with this good range can be found, a laser camera and a long range radar. The laser camera gives a very good image of the approaching car, when dark, foggy weather or snowy weather, or everything all together. The image is displayed for the

42

driver and it is up to him or her to handle the situation. The drawback with the laser camera is, however, that it doesn’t really fill any purpose on a clear sunny day. Also, for the car to handle the situation on its own, advanced image processing is necessary. The long range radar can detect objects up to 200 meters but the resolution at this point is fairly small. As with the laser camera, bad weather condition is not a problem. The radar will not require any advanced image processing and it is as reliable during the day as during the night. The radar can alert the driver with light and sound at a early stage and if he or she still doesn’t react the car can start to brake. Therefore, the best system for this kinds of accident is the long range radar.

5.3.3 Rear end collision This type of accident often takes place in dense traffic at low speeds and therefore not many people are killed. However a large number are injured every year, most of them probably traveling in the car that gets hit from behind. The whiplash injury is very common in these accidents and arises then the head gets slung back and forth. If this type of accident is to be prevented the system must be able to distinguish a normal queue driving and what can actually be a rear end collision. Beside a powerful algorithm the sensor has to get the system accurate and precise information.

5.3.4 Proposed solution for rear end collision In rear end collisions, the speed of the involved cars is rather limited. Therefore, it is not as important that the safety system detects the other car at a early stage. The range of the short range radar is about 30 meters and this would give enough the rear car enough of time. In traffic jam situation, the speed of the cars is often about 20-30 km/h at most. This is equal to 5,56-8,33 m/s. The three seconds of safety limit is still not more than about 25 meters. The resolution and displacement of the short range radar is also very good. It can determine how far away the other car is down to five centimeters. The ability of working in bad weather condition is also very good. Similar to the interception accident, a warning light and sound would activate if an accident is close. As a final option the car would start to brake to decrease the damage. The laser camera can also be a good option, but then again the driver has to monitor the display unless the car has to be equipped with advanced technology for image processing. A camera works fine under these conditions but here the car also needs advanced image processing. The range of ultrasound is not sufficient for these kinds of accident. The best safety system would therefore be the short range radar.

5.3.5 Pedestrian accidents Pedestrians are vulnerable in traffic accidents. Today several car manufacturers are working with safety systems that mitigate the injuries. With an active safety system the accidents can be prevented and there will be no injuries at all. It can be hard for a driver of a car to recognize a person near or on the road, especially then it’s dark. Other cars and vehicles have lamps both in front and rear, but pedestrian seldom use reflex or lamps. A system that helps the driver avoid these accidents has to work in all whether conditions and at night. Also it must be powerful enough to distinguish pedestrians from other objects near the road.

43

5.3.6 Proposed solution for pedestrian accidents It is particularly hard to distinguish a pedestrian from other objects when the car is far away. The system need to know if it is in fact a person and not a tree or a sign post. Because of the long distance required the only options left are the laser camera and long range radar. As mentioned before, the laser camera is very limited to working the best during the night in bad weather condition. The long range radar can detect the pedestrian at a great distance but it might have problem to distinguish whether it is a pedestrian or not. The best solution could be to combine a long range radar and a short range radar. The short distance radar can detect objects down to 15 centimeters wide which is enough for detecting a person. If the long range radar detects something further than the short range radar can reach a light- and sound warning activates. When the car comes closer the short range radar can detect the person and distinguish if he or she is moving towards the cars path or not. If a dangerous situation is about to occur the car would start to brake.

5.3.7 Combined proposition The best way to combine the three scenarios from above is to use a combined Long Range Radar and Short Range Radar. With this solution it is possible to have the great range of the LRR and still the high accuracy and resolution of the SRR. The radar would continuously switch between 24 GHz and 77 GHz. The echo is then picked up by two different antennas, one for each frequency. Using the Laser Camera would be two expensive and demands more advanced computer processing. In the feature, on the other hand, when computers capable of advanced image processing are faster and cheaper, the best solution could be to combine a radar and a camera. The radar to determine the distance to the object and the camera to distinguish what kind of object it is.

6. Manufacturing Strategies This section will cover market- and manufacturing aspects, analyze potential fields in safety system production for where our company can become successful. A small market analysis in table form will be presented to provide a good overview of what companies are active on the present market and provide car manufactures with safety system equipment. In addition, some outlooks and strategies for the future will be provided at the end.

44

6.1 Market Analysis Company Country Employees Safety Products Customer

Delphi US 171 000 Mixed, Remote Sensors All

Valeo France 71 000 Mixed, Park4u All

TRW US 63 100 Mixed, ACC, LDW All

Autoliv Sweden 41 800 Airbags, Remote Sensors All

Continental Germany 24 000 Sensing, ACC, LDW Unkown

Siemens Germany Radar, Lidar, Camera BMW

Iteris US Image Processing, LDW Infiniti

Hella Germany 24 000 ACC, Night Vision

Innosent Germany 40 LDW, ACC

Mobileye Isreal 160 Visual Detection BMW 07, Delphi Table 2 Safety system providers for car industry

It is obvious by looking at the table that safety system manufacturing is a highly competitive field. Giants like Delphi and Autoliv, which serve almost every car manufacturer, have become very successful even though they have slightly different focus on their product portfolio. However, a company can never be satisfied and comfortable with their position. One must constantly develop and adopt new technology.

6.2 Potential Fields for Production Many of the fields for production of safety system are already heavily crowded and subjected to tough competition by many big corporations. Therefore, areas like radar production might not look very promising at the moment. However, there is always a need for development and a radar product that combines long and short radar capabilities at a low price would certainly be interesting for the market. In addition, the field of visual detection equipment is still quite small. This field looks very promising and with the right strategy there are good chances to become successful.

6.3 The Future There is no doubt that the area of safety system and detection equipment will continue to develop and provide market opportunities. The most promising of them all at the moment looks to be visual detection with cameras. A camera can provide much more information in a picture than radar sensors could ever do. And it is likely that in the future, cameras together with computers will have the same ability to judge distance as the human eye making radar obsolete. To sum up, a smart move for the future would be to start development and production of visual detection equipment.

45

7. Discussion Active safety systems, as every type of system is effected by malfunctioning, these can be classified into False warnings and Lack of warnings. This means that we deal with a responsibility question if the system causes an accident. The responsibility of the car behaviour in every condition is the drivers and he must have a careful approach in respect to him and to other road users. Maybe that a safety system is guaranteed by the car maker, in this case the responsibility is an important aspect. The car manufacturer have to make his system very well in order to avoid malfunctioning, otherwise the brand image will go down. Other difficulties in introducing safety systems are due to laws and regulation. In the U.S.A. for example, if a car maker introduces new safety systems or improve the safety of its cars, the others competitors has to follow it. If a car maker is sued it has to demonstrate that it manufactured its product as well as possible. This is a constraint especially for Volvo whose core value is safety and is owned by Ford. If Volvo improves car safety e.g. a new kind of roof, it will be a problem for Ford. The problem is that Ford has to follow Volvo solution in order to not get sued. This means that Volvo can build safe cars but not promote every feature. This problem can be solved by government support. If the primary aim of a car maker is to sell its vehicles, and it can use active safety systems to differentiate their products, in order to increase market penetration; on the other side governments have the main objective to reduce accidents (save life at first, “at second” the indirect costs are about 160 billion euro/year, equivalent to 2% of the EU’s GNP). We can see that there are different objectives between car makers and governments, but there is a common starting point, accidents avoidance. Europe has several large-scale programs in progress under the umbrella of Road Transport Informatics (RTI), which is the equivalent of the U.S. ITS. Their main programs are dedicated road infrastructures for vehicle safety in Europe (DRIVE) and the program for European traffic with highest efficiency and unprecedented safety (PROMETHEUS). While the projects are separate, close cooperation between the two is needed to reach a common goal. Improvements in safety can be done acting directly on the car and acting on the infrastructures e.g. wireless communications between car and road (see sec 3). Another step in reaching the objectives of both car manufacturers and government may be to have discounts or particular economic treatments on “safety cars” i.e. discounts on insurance for the driver or less taxes for the car manufacturer. If a government puts in money and subsidize a safety measure or demands a safety system trough legislation, even systems that are expensive and not profitable can be introduced on the market.

46

Refrences [1] Japan External Trade Organization http://www.jetro.go.jp/en/market/attract/automotive/safety.html [2] Continental http://www.conti-online.com/generator/www/de/en/cas/cas/themes/products/ electronic_brake_and_safety_systems/introduction_ebss_en.html [3] Mitsubishi Motors http://www.mitsubishi-motors.com/corporate/about_us/technology/safety/e/abs.html [4] Lexus http://www.lexus-europe.com/technology_explorer [5] BMW http://www.bmw.co.za/Products/FIRST/Active/ [6] Mazda http://www.mazda.com/csr/custormer/04-02.html [7] “Federal Motor Vehicle Standards; Electronic Stability Control Systems” National Highway Traffic Safety Administration, the U.S Department of Transportation Docket No. NHTSA–2006-25801, RIN: 2127-AJ77 [8] Toyota http://www.toyota.co.jp/en/safety_presen/tech [9] Honda safety [10] Chapuis, R. et al., Accurate Vision Based Road Tracker, in Proceedings of Intelligent Vehicle Symposium, Versailles, 2002. [11] Saito, A., Kimachi, M., and Ogata, S., Silhouette Vision: New Video Vehicle Detection Field Proven Robust and Accurate, in Proceedings of 6th World Congress on Intelligent Transport Systems, Toronto, November 1999. [12] Faugeras, O., 3D Reconstruction of Urban Scenes from Sequences of Images, Research Report INRIA RR-2572, June 1995. [13]Broggi, A., Bertozzi, M., and Fascioli, A., Self-Calibration of a Stereo Vision System for Automotive Applications, in Proceedings of IEEE International Conference on Robotics and Automation, Seoul, Korea, 2001. [14] Zhang, Z. et al., A Robust Technique for Matching Two Uncalibrated Images Through the Recovery of the Unkown Epipolar Geometry, Research Report INRIA RR-2273, May 1994. [15] BMW www.bmw.com

47

[16] Mercedes www.mercedes.com [17] Press Release DaimlerChrysler, ”The new Mercedes-Benz S-Class: Thinking ahead - setting the pace”, April 2005. [18] I. Gresham, A. Jenkins et al., ”Ultra-wideband radar sensors for short range vehicular applications”, IEEE Trans. Microwave Theory and Techniques, Vol. 52, N0. 9, pp. 2105-2120, Sept. 2004. [19] S. Lakshmanan, K. Kluge, ”LOIS: A real-time lane detection algorithm” Proc. Conf. Information Sciences and Systems, Princeton, NJ, pp. 1007-1012, 1997. [20] S. K. Kenue, ”LANELOK: Detection of lane boundaries and vehicle tracking using image processing techniques - Part I and II” SPIE Mobile Robots IV, Vol. 1195, 1989. [21] T. Emura, M. Kumagai, and L. Wang A Next-Generation Intelligent Car for Safe Drive, Journal of Robotics and Mechatronics, vol. 12. No. 5, pp.545-551, 2000. [22] K. Jurgen et al., Object Detection, Collision Warning and Avoidance Systems, Automotive Electronics Series, Society of Automotive Engineers, Inc, 1998. [23] “Anaysis of Fatal Motor Vehicle Traffic Crashes and Fatalities at Intersection, 1997 to 2004”, U.S Department of Transportation, National Hihway Traffic Safety Administration (NHTSA), DOT HS 810 682 http://www.nhtsa.dot.gov/portal/site/nhtsa/menuitem.6a6eaf83cf719ad24ec86e10dba046a0/ [24] "Traffic Safety Facts 2005" U.S Department of Transportation, National Hihway Traffic Safety Administration (NHTSA) http://www.nhtsa.dot.gov/portal/site/nhtsa/menuitem.6a6eaf83cf719ad24ec86e10dba046a0/ [25] Citroen. “Lane Departure Warning System – LDWS”. 2007-05-16. http://www.citroen.com/CWW/en-US/TECHNOLOGIES/SECURITY/AFIL/AFIL.htm [26] Volvo. “Geneva 2005 – Volvo Cars focuses on preventive safety”. 2007-05-16. http://www.volvocars.com/corporation/NewsEvents/News/news.htm?item=%7B6B5A7DC7-10F5-425F-B3F6-009650F27E34%7D [27] “Cooperatively controlled collision avoidance” Zu Zhang – Brookhaven National Laboratory, Upton NY 119773 [28] http://www.uktelematicsonline.co.uk/index.html [29] http://www.uktelematicsonline.co.uk/html/wireless_wan_s.html [30] http://www.toyota.co.jp/en/tech/its/future/development.html [31] http://www.toyota.co.jp/en/tech/its/future/tie_up3.html [32] http://www.tele.soumu.go.jp/e/system/ml/its/details/future.htm

Mikael Ljung lecture in Active safety, Chalmers University of Technology, 2007-04-23 Baddeley, A.D. &Hitch, G.J. (19974). Working Memory. In G. Bower (Ed.), The Psychology of Learning and Motivation (pp. 47-90). New York: Academic press Shinar, D. (1978). Psychology on the road. The human factor in traffic safety. John Wiley&sons, New York

Appendix A The main goal of an active safety system is protect the driver from hazard situation that he can occur during driving. This goal is achieved by the problem definition (what is the situation that the system must avoid), the environmental factors ( the external factors that interface with the car and driver) and the driver behaviour ( in which way the system and the environmental factors influence the human behaviour). To define and explore which factors and interactions influence the origin and development of traffic accidents need to develop accident investigation methodology for active safety development; identify factors that contributes to the occurrence of accidents; eliminate the factors with appropriate countermeasures, i.e. technological functions. To identify the factors that occur in the accidents, the driver behaviour model must be considered in way to use the human factors as control. In this concept, driving is about controlling a process and it can do in order to both understand: how well the driver control the situation (degree of control); and on which level the control is being carried out. Contextual Control Model (COCOM) and Extended Control Model (ECOM) are used to define the term of controls, respectively[99].

Then the involvement of human factor in accidents must be analysed in order to investigate how the result parameter of the driver behaviour model, in the concept of driving as control process, influence the hazard situation. Driving Reliability and Error Analisys Method (DREAM) is used to achieve this objectives.

COCOM model

ECOM model

The building blocks for dream are the MTO – model and the Joint Cognitive Systems. The MTO – model describe the interaction between Man, Technology and Organization as the environmental road structure. While the Joint Cognitive System describe the ensemble of vehicle (artificial cognitive system) and man (natural cognitive system) as a system that interface with the external factors in the same way. DREAM uses three tools: a system to describe the Common Performance Conditions (CPC) that affects all drivers regardless (weather, light, etc); a categorisation system which lists possible causes and consequences in an accident/incident event; a step-by-step procedure description for how to perform the analysis.[99] The human behaviour are strongly related with the country culture that can see the active safety system in several way. A theory to analyze many factors which influenced the human behaviour is the Theory of Planned Behaviour (TPB). According to TPB people’s attitude towards the behaviour, their subjective norm (similar to peer pressure) and their perceived behavioural control determine their behaviour indirectly via their intention . People’s attitude towards a behaviour is determined by their beliefs about the likely consequences of the behaviour, their subjective norm is determined by their beliefs about what important others think of the behaviour, and their perceived behavioural control is determined by their beliefs about factors that may facilitate or obstruct the performance of the behaviour. The intention is defined as a willingness to try to perform the behaviour and the behaviour refers to an action[89]. [99] Mikael Ljung, Active safety -How to define the problems, and how to create the solutions- Ppt presentation Crash Safety Division Chalmers, http://www.mvs.chalmers.se/~mys/ActiveSafety07.htm [89]Henriette Wallén Warner –Factors influencing drivers´ speeding behaviour- ppt presentation http://www.mvs.chalmers.se/~mys/ActiveSafety07.htm

Theory of Planned BehaviourTheory of Planned Behaviour(TPB)(TPB)

BehaviourBehaviourNormativa Normativa BeliefsBeliefs

SubjectiveSubjectiveNormNorm IntentionIntention

PerceivedPerceivedBehaviouralBehavioural

ControlControl

AttitudeAttitude

ControlControlBeliefsBeliefs

BehaviouralBehaviouralBeliefsBeliefs

((AjzenAjzen, 1991), 1991)