1. Final EU-MOP project workshop EU-MOP Intelligent Design
Dennis Fritsch, Fraunhofer IPA
Madrid, January 2008

2.   Introduction Why is a swarm better than a single unit? Swarm
Scalability
Fault tolerance
Multi
locality
Flexibility
System performance  

3. Introduction
In order to use these advantages, the units have to be intelligent.
This artificial intelligence will be achieved by several sensors and a control system:
Plan
Act
Environment
Sense

4. Introduction

5. Introduction

6. Sensor systems Sensors will be mainly needed for navigation
Where am I?
Position
Orientation
Where do I want to go?
Oil
Internal state
How can I get there?
Other robots
Environment

7. Sensor systems Proceeding Classification of sensors
Identification of sensor systems
Test of sensor systems
Definition of sensor configurations

8. Sensor systems
Step 1: Sensor classification matrix

9. Sensor systems
Step 2: Identification of sensors

10. Sensor systems
Step 2: Identification of sensors

11. Sensor systems
Step 2: Identification of sensors

12. Sensor systems
Step 2: Identification of sensors

13. Sensor systems
Step 3: Test of sensors

14. Sensor systems
Step 4: Definition of several sensor configurations, e.g.:
Sensor system
L model
S model
Oil sensor
@ MS 
DGPS
High end
Low cost
IMU 
RADAR
Depth sensor
Bumpers
Fuel tank / oil storage sensors
Radio system

15. Control / Swarm strategies
6 basic operations …

16. Control / Swarm strategies
… lead to 11 basic tasks:
move to position (x,y)
avoid collision (pre-collision)
react to collision (post-collision)
move to known oil position
skim oil and stay in oil slick as long as possible (information)
chaotic operation
search oil
skim oil and stay in oil slick as long as possible (oil sensor)
return to MS / collection point
move into operational area
stop engine and wait for new commands

17. Control / Swarm strategies
Strategies for task move to position (x,y)

18. Control / Swarm strategies
Strategies for task avoid collision (pre-collision)

19. Control / Swarm strategies
Strategies for task skim oil and stay in oil slick

20. Control / Swarm strategies
Strategies for task chaotic operation

21. Control / Swarm strategies
Strategies for task move into operational area

22. Robot simulation

23. Robot simulation Input GUI Output Oil fate model by OXF
Analysis by BMT
Parameter file
Oil fate model

24. Robot simulation
Real case scenarios Sea Empress Oil Spill at Tenby Harbour

25. Robot simulation
Results:
Identification of
best sensor configurations
best swarm strategies
But:
Simulation is like masturbation. If you do it too often, you begin to think it is the real thing.

26. Development of Concept
Proof of concept
Proceeding
Development of Concept
Hardware
Robots
Projector
Camera …
Software
Camara: Robot localisation
Projector: Oil spill projection

27. Proof of concept
Concept

28. Proof of concept
Hardware Components

29. Proof of concept
Hardware Components

30. Proof of concept
Hardware Components

31. Proof of concept
Software for oil projection

32. Proof of concept
Software for oil projection

33. Proof of concept
Software for oil projection

34. Proof of concept
Programming of robots

35. Proof of concept
Programming of robots

36. Proof of concept
Video

37. Proof of concept
Video

38. Proof of concept
Video

39. Proof of concept
Video

40. Results