cps verification cps week.pptable/acps2011/presentations/7_butts.pdf · verification of...
TRANSCRIPT
Toyota’s Direction
Infrastructure integration
Our sustainable mobility strategy Our sustainable mobility strategy includes products, partnerships, the includes products, partnerships, the urban environment and energyurban environment and energyurban environment and energy urban environment and energy solutions.*solutions.*
*Toyota 2010 North American Environmental Report Highlights*Toyota 2010 North American Environmental Report Highlights
Toyota Technical CenterPowertrain Control – Model based Development
1Vehicle technologies
Toyota 2010 North American Environmental Report HighlightsToyota 2010 North American Environmental Report Highlights
Presentation FlowPresentation Flow
1. Development Challenges1. Development Challenges2. Focus on Verification and Validation3 Open Challenge: Engineering Processes for3. Open Challenge: Engineering Processes for
Verification and Validation of Cyber-Physical Systemsy
CPS Week 2011Toyota Technical Center
Powertrain Control – Model based Development
2
Product Direction: Today’s Complexity - system scale
- variations
Environment
LEXUS LS460 …• 100 Electronic Control Units (ECUs) or more• Program size of seven million lines
catio
n
Large-scale and Comp
Engine control
Auto cruise controlAir conditioner control
Nikkei business (September, 2006)
umbe
r of
pag
es o
f ng
ine
cont
rol s
peci
fic
Transmission control
Vehicle integration control
Suspension control
Power steering control
S f t
1988 1997 2002 2005 2007
nu en
Hybrid control
Suspension control
Brake control
Electric power control
AWD control
Safety Comfort
Control, integration, and complexity are accelerating in order to improve vehicle f d id f t
In-vehicle control systems
Toyota Technical CenterPowertrain Control – Model based Development
performance and provide new features.
3
Product Direction: Tomorrow’s Complexity: Cyber-Physical Systems- Vehicle to Infrastructure and Vehicle to Vehicle
C lli i A id d C ti D i i- Collision Avoidance and Cooperative Driving I2VInfra struc ture
V 2VC oop erat ion
P ath P lannin g
B ehav ior
N o m inal Path M ap
V 2VS afety M app in g
R eg ulat ion L oca liz at ion
C oll ision-F ree P ath
M ap
Veh ic le St a te Ob sta cles
P hy s ic al
G P S , R ada r, C am er a
A ct uator M ove m en ts
M eas urem e nts
CPS: distributed computation nodes operating in a control hierarchy that interact ith th hi l ’ h i l i l ti t id iti l f ti
V ehicle 3 C ontro l le r
Veh ic le 2 C on tro ller
V ehicle 1 C ontro l le r
Toyota Technical CenterPowertrain Control – Model based Development
with the vehicle’s physical processes in real-time to provide critical function.
4
Today’s Situation: Control Development Process OverviewToday s Situation: Control Development Process Overview
Insufficient requirement analysis and test designHigh possibility of re-design and re-evaluation
Focus on right hand side of V-processInefficient V&V due to exhaustive experiment with the actual product
Huge number of tests and more difficult evaluation
Market ProductNew
knowledge
gApproaching the limits of human power
Packagetest
needs strategyknowledge
Changedtarget
Req. analysis
System design
Control designPrototypingsystem
Req.analysis Evaluation
Spec C-code
Mass Production Phase Advanced Phase Very Advanced Phase
MassProduction
Small multiple V-processes are executed on big multiple V-processes
Toyota Technical CenterPowertrain Control – Model based Development
5
Consequence of Development Complexity
H
Getting larger and more complex control system
Development in shorter time
Heavy workload
Globalization,Fuel economy
Lack of researchLack of research for new development
method or process
Conventional development Low efficiency
A shift in paradigm is needed to overcome the above “Demon Cycle”Toyota has been pushing ahead with MBD as an essential solution.
Toyota Technical CenterPowertrain Control – Model based Development
Toyota has been pushing ahead with MBD as an essential solution.
6
Model-Based Development: Basic Tooling
Virtual WorldValidation
Control SoftwareSpecification=
Engine PerformanceSpecification=
Plant Model Controller ModelSILS / MILS
Combination
Plant Model Controller Model
HILSRapid Prot. ECU
Plant(Engine, Transmission etc.)
Controller(Hardware, Software)
Real World
Combination
Validation
Toyota Technical CenterPowertrain Control – Model based Development
Real World
7
MBD Focus areas
1.1. Process and Information ManagementProcess and Information Management
22 Plant ModelingPlant Modeling2.2. Plant ModelingPlant Modeling
3.3. ModelModel--based control designbased control design
4.4. CalibrationCalibration4.4. CalibrationCalibration
5.5. Verification and ValidationVerification and Validation
Toyota Technical CenterPowertrain Control – Model based Development
8
Verification and ValidationSt t- Strategy -
1. V&V process(ex. Hierarchical verification, Requirements specification)
2. Verification with advanced techniques(ex Automated test generation Formal methods)
Req. analysis
System design Calibrationand Test
Control systemreq. spec Product
(ex. Automated test generation, Formal methods)
3. Efficient Validation including experiment(ex. Software structural analysis, Defect cause exploration)
Control design Integration
Control systemdesign spec
Control softreq. spec
ECU
and Test
( y , p )
4. V&V environment(ex. Integrated V&V environment, Test-reuse, Test Automation Data Base) Code gen.
Control softdesign spec
C-code
Control design Integration
Renew development process by applying advanced V&V technologies to improve: Quality - no defects allowed in product Efficiency - minimized development cost
Toyota Technical CenterPowertrain Control – Model based Development
Efficiency minimized development cost
9
Verification and Validation- Needs and technologies -
CPSCPS
Vehicle
Problem scaleLine Off
bl lit
CPSCPS
HILS
Unit
- able quality
No hesitation in coldenvironment
O tp t e o of the
System
(Closed-loop)
SILS
Gap
Output error of thesystem shall be lessthan 5%
Needs found in the actual development
Feature Output error of the featureF l th d
Automated test generation
Existing “advanced” technologies
ModuleVagueness ofrequirement
Output error of the feature shall be less than 5%
Output of the function is always in certain range
Formal methodsSoftware
Toyota Technical CenterPowertrain Control – Model based Development
Quantitative/Static(ex. Logic)
Quantitative/Dynamic(ex. Trajectory)
Qualitative/Dynamic(ex. hunting)
Sensibility(Quickness)
10
Verification of Cyber-physical systems (e.g. collision avoidance)• The systems are vastly heterogeneous in nature. What mathematical or modeling
t ti h ld b d t d it lid ti d ifi ti ?representations should be used to admit validation and verification?• The systems are complex. What is the maximum complexity that can be validated and
verified? How do we quantify complexity? • Given detailed design models of the system components (these are the most readily available g y p ( y
in today’s processes) how does the engineer properly abstract the models to satisfy the complexity constraint?
• The system functionality is hierarchically structured. Are there validation and verification techniques that accommodate (and possibly exploit) this structure?techniques that accommodate (and possibly exploit) this structure?
• The systems are comprised of outside elements (e.g. infrastructure, unknown vehicles) that cannot be brought in-house for analysis or testing. How do we parameterize the outside element’s performance envelope for validation and verification?The s stems rel on loss ireless comm nication Ho do e characteri e the orst case• The systems rely on lossy wireless communication. How do we characterize the worst-case scenarios so that we can validate and verify the dynamic behavior of the system?
• This system configuration is multi-agent and dynamic. How do we validate and verify given the untold number of agent scenarios?
• The systems are subject to malicious interaction. How do we validate and verify the system’s countermeasures?
• The vehicle performance envelopes are subject to several uncertain parameters such as weight, road coefficient, road grade, tire wear. Are there validation and verification techniques
Toyota Technical CenterPowertrain Control – Model based Development
weight, road coefficient, road grade, tire wear. Are there validation and verification techniques able to accommodate these uncertainties?
11
Open Challenge: Engineering Processes for Verification and Validation of Cyber-Physical Systemsy y y
a) We propose a community challenge to:a) describe what it means to validate and verify cyber-
physical systems in concrete terms,b) develop engineering processes that practicing engineers
can use to conduct the validation and verification, and) t k b bli hi t i th l dc) track progress by publishing metrics on the scale and
complexity of systems than can be validated and verified using the processes.
b) Collision avoidance systems could be used as a representative application test-bed:a) much of the work is already government sponsored anda) much of the work is already government sponsored, andb) the benefits are societal.
Toyota Technical CenterPowertrain Control – Model based Development
12
Perhaps we need to generate a Moore’s Law for Validation and Verification of Cyber-Physical Systems ?
Thank youThank you
Toyota Technical CenterPowertrain Control – Model based Development
13