1. KTH ROYAL INSTITUTE
OF TECHNOLOGYMulti agent control
for cooperative coverage
Candidate
Mario Sposato
Examiners
Prof. Karl Henrik Johansson
Prof. Dimos Dimarogonas
Supervisor
Antonio Adaldo
Stockholm, Sweden
February 25th, 2016

2. Overview
● Introduction
● Modelling and control
● Path Planning
● Coverage problem
● Experiments
● Conclusions and future developments.

3. Introduction
● Mission: autonomous
deployment of a team of
aerial robots to collect
information over a given
environment.
● Networked UAVs used as
mobile sensors.

4. Modelling

5. Control design
● Problem: Regulation to a point and trajectory tracking:
where is a given reference trajectory.
● Possible solution: backstepping procedure
+

6. Control design: Backstepping
● Finding a Candidate Lyapunov Function and designing a
control input to stabilize to zero.
● Three steps of the backstepping procedure give:
1.
2.
3.

7. Control design: yaw control
● Track a reference trajectory for the yaw angle:
● We choose the controller

8. Control design: simulations

9. Control design: simulations

10. Path planning: potential fields
● Executed trajectories become smoother.
● Collision avoidance techniques can be implemented.

11. Path planning : Attractive term

12. Path planning : Collision avoidance
● Pushes the quadcopters away from each other when the
distance is below a threshold ς and they are approaching
each other.

13. Experiment: Collision avoidance
● A stick is used to represent a mobile obstacle.
● It can be seen that the quad is repelled by the stick.

14. Coverage problem: visibility function
● Objective: Deploy a set
of mobile sensors to
enhance the perception
of a given environment.

15. Coverage problem: Abstraction

16. Coverage problem: formulation

17. Coverage problem: Voronoi configuration
● Definition: the system is said to be in
a Voronoi configuration with
tolerance ς > 0 if:
● Lemma: if a configuration (A,Q) is
optimal, then it is also Voronoi.

18. Coverage problem: Proposed approach
Landmarks transferring

19. Coverage problem: Proposed approach
Pose optimization

20. Coverage problem: distributed implementation

21. Coverage problem: distributed implementation

22. Coverage problem: distributed implementation

23. Coverage problem: distributed implementation

24. Comprehensive experiment

25. Conclusions and future developments
● Conclusion:
○ Model and Controller for the quadcopter.
○ Path planning with collision avoidance.
○ Coverage mission.
● Future development:
○ 3D coverage.
○ Cooperative structures inspection.

26. Thank you! - Questions?
● Thank you for your attention!
● Questions?