In this paper, we present an optimal sensor management technique for an Unmanned Aerial Vehicle (UAV) to autonomously geo-localize multiple mobile ground targets. The target states are continuously estimated using target locations asynchronously captured by a gimbaled camera with a limited field of view and processed with a set of Extended Kalman Filters (EKFs). The technique incorporates a Dynamic Weighted Graph (DWG) method to first group estimated targets and then determine regions with high target densities. A Model Predictive Control (MPC) method is used to compute a camera pose that minimizes the overall uncertainty of the target state estimates. The validity of the proposed technique is demonstrated using simulation results.

1. Optimal Sensor Management
Technique For An Unmanned Aerial
Vehicle Tracking Multiple Mobile
Ground Targets
Negar Farmani, Liang Sun, Daniel Pack
Unmanned System Lab
The University of Texas at San Antonio

2. Outline:
– Introduction
– Problem Statement
– Optimal Sensor Management
• Coordinates Transformations
• Gimbal Pose Selection
– Experimental Results
– Conclusion

3. Introduction
– Tracking multiple mobile targets in optimal
manner
– Previous works
– Novel optimal sensor management method

4. Outline:
– Introduction
– Problem Statement
– Sensor Management
– Experimental Results
– Conclusion

5. SAMPLE SCENARIO

6. Outline:
– Introduction
– Problem Statement
– Sensor Management
– Experimental Results
– Conclusion

7. UAV And Target Relation

8. Associated Coordinate Frames
• Three frame of interest:
Body Frame = (ib , jb , kb)
Gimbal Frame= (ig , jg , kg)
Camera Frame= (ic , jc , kc)

9. Target Projection

10. Target Geo-Localization

11. Estimation Technique
Extended Kalman Filter (EKF)
– Motion model:
– State of target:
– Measurement model:

12. Geo-localization
Extended Kalman Filter (EKF):
– Prediction Step:
where

13. Geo-localization
– Measurement Update:
where

14. Candidate Generation for Gimbal Pose
• Dynamic weighted graph of targets

15. Candidate Generation for Gimbal Pose
• Check FOV
– Limited FOV
– Camera gimbal pointing direction

16. Target
Estimations
Gimbal
Candidates
MPC
Technique
Target
Estimations
DWG FOV Test
Gimbal
Candidates
MPC
Technique
Sharma & Pack method:
Proposed method:

17. Sharma & Pack Method
• R. Sharma and D. Pack, “Cooperative Sensor Resource Management for Multi Target
Geo-localization using Small Fixed-wing Unmanned Aerial Vehicles.” in Proc. AIAA
Guidance, Navigation, and Control (GNC) Conference, American Institute of
Aeronautics and Astronautics, 2013.
– Develop a vision based cooperative sensor fusion technique to geo-locate
multiple mobile ground targets using
– Develop a cooperative sensor resource manager using Model Predictive
Control

18. Candidate Generation for Gimbal Pose
• Model Predictive Control(MPC)

19. Outline:
– Introduction
– Problem Statement
– Sensor Management
– Experimental Results
– Conclusion

20. Experimental Results

21. Experimental Results

22. Experimental Results

23. Experimental Results
AVERAGE GEO-LOCATION ERRORS OF FIVE TARGETS FOR THE 100
EXPERIMENTS USING THE PROPOSED METHOD AND THE ONE REPORTED
IN [9].
Target No. 1 2 3 4 5
North Position(m) 15.49 10.73 24.83 24.46 15.26
North Position(m)[9] 11.78 12.54 26.32 31.9 23.12
East Position(m) 13.58 6.08 14.28 8.71 7.6
East Position(m)[9] 9.69 10.28 16.38 13.7 11.41

24. Experimental Results
Overall Error Error
(m)
Improvement
(%)
Overall Error in North Position (m) Overall
Error in North Position (m)[9]
18.15
21.14
14
Overall Error in East Position (m)
Overall Error east Position (m)[9]
9.94
11.91
16
OVERALL AVERAGE OF GEO-LOCATION ERRORS

25. Experimental Results

26. Conclusion And Future Works
• A new sensor management technique for UAVs tracking
multiple targets
– A dynamic weighted graph
– A Model Predictive Control technique
• Future Work
– Multiple UAVs cooperatively tracking multiple targets