feup asasf– unmanned air vehicles project group developing of a softwall controller to avoid...

30
FEUP AsasF– Unmanned Air Vehicles Project Group Developing of a Softwall Controller to avoid No-Fly-Zones in an existing Autopilot System Christoph Bruno Ruetz Departamento Engenharia Electrotécnica e Computadores Faculdade de Engenharia da Universidade do Porto R. Dr. Roberto Frias, 4200 465 Porto, Portugal E-mail: [email protected]

Post on 20-Dec-2015

213 views

Category:

Documents


0 download

TRANSCRIPT

FEUP

AsasF– Unmanned Air Vehicles Project Group

Developing of a Softwall Controller to avoid No-Fly-Zones in an existing Autopilot System

Christoph Bruno Ruetz

Departamento Engenharia Electrotécnica e ComputadoresFaculdade de Engenharia da Universidade do Porto

R. Dr. Roberto Frias, 4200 465 Porto, PortugalE-mail: [email protected]

2AsasF– Unmanned Air Vehicles Project Group

FEUP

Structure

Motivation

Apollo System

Softwall System• Goals• Requirements

Approaches

Implemented System• Goals• Software structure • Problems

Future Developments

3AsasF– Unmanned Air Vehicles Project Group

FEUP

Motivation

After 9/11, more than ever safety in and around air vehicles is needed

Ever since, several ideas appeared:• to forecast if a plane will fly into a building

• to prevent a pilot from flying into a building

Control Commands should not be directly sent to the airplane steering

• to design a System, it has to be between the control and steering

4AsasF– Unmanned Air Vehicles Project Group

FEUP

Apollo System

UAV controller component of the AsasF control hierarchy, developed in FEUP

Decouples detailsof UAV control fromorganization ofexternal controllers

5AsasF– Unmanned Air Vehicles Project Group

FEUP

Softwall System [Goals]

Define a system that avoids detected Softwalls

The System should be a part of Apollo Modularity

But anyway there should be a possibility to decide whether the module will be used or not

6AsasF– Unmanned Air Vehicles Project Group

FEUP

Softwall System [Requirements / Environment]

Language was C++, because Apollo is written in C++

modularity good planning of Object structure

SVN for developing both projects (Apollo and Softwall) at the same time

Defining softwalls

Algorithms that meet requirements to detect and avoid softwalls

7AsasF– Unmanned Air Vehicles Project Group

FEUP

Approaches (I)

Reachability Sets from Ian Mitchell [1]• Backward calculation from a obstacle from where it‘s

possible to reach it

• heavy calculations

• Existing matlab sources but no c++ giving up this approach

8AsasF– Unmanned Air Vehicles Project Group

FEUP

Approaches (II)

2D-Softwall graphical approach from J. Adam Cataldo [2]

• Base: Critical measurement 1/T

Depends on the direction, the distance and the max turning rate of the airplane

Safe!!! Tested in a Research Project [3]

9AsasF– Unmanned Air Vehicles Project Group

FEUP

Approaches (III)

Using Airplane Model

s := speed; := heading; t := time; p := position

10AsasF– Unmanned Air Vehicles Project Group

FEUP

Approaches (IV)

How does it work…?

No-Fly-Zone

radius

collisionAvoid window Waypoint

collision

collision

11AsasF– Unmanned Air Vehicles Project Group

FEUP

Approaches (V)

Checking the critical time every 100 ms

In an avoiding window…• the pilot will be informed about a potential collision

• starts at 4th and ends at 6th times of the min turn radius

If missing, evasive action with the max. turn rate initiates

After a certain time the airplane tries to follow the old plan

Checking begins again

Like a POTENTIAL FIELD ALGORITHM

12AsasF– Unmanned Air Vehicles Project Group

FEUP

Approaches (VI)

This approach is the basis of the Avoiding system Problems: Calculation only during the flight

not smoothed enough Solution: pre-calculation of a possible trajectory as a

second algorithm set upon the first algorithm

Trajectory calculation is needed• A lot of different kind of algorithms from Industrial

robot sector, like…• Roadmaps (Voronoi, Visibility Graphs, Decomposition…)

(just good for 2D Environments • A* (discretize the C-Space) • PRMs (unpredictable) • …

Baginsky algorithm

13AsasF– Unmanned Air Vehicles Project Group

FEUP

Approaches (VII)

Why Baginsky?• Geometrical Approach from Boris Baginsky

• Heuristic

• Short search time in low dimension <= 3D

• Good path quality

• Good solutions in 2D or 3D environments

How does it work…?

14AsasF– Unmanned Air Vehicles Project Group

FEUP

No-Fly-Zone

radius

Waypoint

15AsasF– Unmanned Air Vehicles Project Group

FEUP

No-Fly-Zone

radius

Waypoint

16AsasF– Unmanned Air Vehicles Project Group

FEUP

No-Fly-Zone

radius

Waypoint

Deepest intrusion point

17AsasF– Unmanned Air Vehicles Project Group

FEUP

No-Fly-Zone

radius

Waypoint

Deepest intrusion point

Middlepoint to move

18AsasF– Unmanned Air Vehicles Project Group

FEUP

No-Fly-Zone

radius

Waypoint

Middlepoint to move

Mo

vin

g t

-tim

es

Collision

19AsasF– Unmanned Air Vehicles Project Group

FEUP

No-Fly-Zone

radius

Waypoint

Middlepoint to move

Mo

vin

g t

-tim

es

No Collision Collision

Collision

20AsasF– Unmanned Air Vehicles Project Group

FEUP

No-Fly-Zone

radius

Waypoint

Middlepoint to move

Mo

vin

g t

-tim

es

No Collision Collision

Collision

21AsasF– Unmanned Air Vehicles Project Group

FEUP

No-Fly-Zone

radius

Waypoint

Middlepoint to move

Mo

vin

g t

-tim

es

No Collision Collision

Collision

22AsasF– Unmanned Air Vehicles Project Group

FEUP

No-Fly-Zone

radius

Waypoint

Middlepoint to move

Mo

vin

g t

-tim

es

No Collision

Collision

Collision

23AsasF– Unmanned Air Vehicles Project Group

FEUP

No-Fly-Zone

radius

Waypoint

Middlepoint to move

Mo

vin

g t

-tim

es

No Collision

Collision

Collision

24AsasF– Unmanned Air Vehicles Project Group

FEUP

No-Fly-Zone

radius

Waypoint

Middlepoint to move

Mo

vin

g t

-tim

es

No Collision

Collision

Collision

28AsasF– Unmanned Air Vehicles Project Group

FEUP

Implemented System [Goals]

Apollo decides if the Softwall System will be in use

• Just activate the module in Apollo or not

Apollo and Softwalls should have access to the same information of the Airplanes

Softwalls and Apollo should work as a module of its own.

• Adapter Classes docking on the module to communicate with the environment

That will guarantee independence by changing Autopilot Systems

29AsasF– Unmanned Air Vehicles Project Group

FEUP

Implemented System [Architecture]

2D-Approach

Baginsky

Generic Autopilot Interface

Softwall Filter System

Apollo

Pathplanner

CollisionManager

AirplaneSoftwallConfig File

Config File

30AsasF– Unmanned Air Vehicles Project Group

FEUP

Implemented System [Software] Define softwall

• Neptus - Config file - SoftwallS

[Circle 1] alt = 100 ;altidude in meters lon = 0.6565 ;longitude in radians lat = -2.1342 ;latitidue in radians radius = 100 ;in meters if should be a circle [Halfspace 1] alt = 100 ;altidude in meters lon = 0.6565 ;longitude in radians lat = -2.1342 ;latitidue in radians heading = 3.159 ;in radians, defines the direction of the

;halfspace if it‘s a halfspace

31AsasF– Unmanned Air Vehicles Project Group

FEUP

Implemented System [Problems] Reachabillity Sets not practicable 2D

Approach

Apollo modularity was changed

Execution of Apollo had to be stopped if the airplane was following a trajectory

32AsasF– Unmanned Air Vehicles Project Group

FEUP

Future developments

Reachability sets translate in C++ code

3D Softwalls calculation

Define airplanes as moving softwalls currently there are only static softwalls

Implement softwalls in other Vehicles

Safety and a real environment should be tested

Testing intersection Halfspaces

Make the plane follow a softwall

Send all calculated points at the same time

FEUP

AsasF– Unmanned Air Vehicles Project Group

Developing of a Softwall Controller to avoid No-Fly-Zone in an existing

Autopilot System

Christoph Bruno Ruetz

Departamento Engenharia Electrotécnica e ComputadoresFaculdade de Engenharia da Universidade do Porto

R. Dr. Roberto Frias, 4200 465 Porto, PortugalE-mail: [email protected]

Obrigado!