Event
Research on Multi-UAV Cooperative Corridor Information Planning
1 reports · 1 sources
Lars Oerlemans, Moji Shi, and Marija Popovic propose a multi-UAV cooperative sensing scheduling method for unknown threat areas, aimed at improving the navigation safety of unmanned ground vehicles (UGVs). The method uses a risk-averse shared exposure belief to guide UGV path planning, and employs task-conditioned rewards to drive UAVs to focus on uncertainty regions related to exposure; UAV observations update the belief online, and the UGV accordingly avoids newly discovered threats. In multi-UAV scenarios, region-conditioned spatial allocation reduces redundant sensing. The authors report that in simulation, compared with terrain-semantic replanning, this method reduces cumulative UGV exposure by 38%, and the multi-UAV coordination scheme reduces redundant aerial coverage from 38.8% to 3.7%.
Report timeline
- UGV-Conditioned Multi-UAV Informative Planning on a Shared Exposure Belief
Simulation results show that the method reduces cumulative UGV exposure by 38% compared with terrain-semantic replanning, and the multi-UAV coordination scheme cuts redundant aerial coverage from 38.8% to 3.7%. The method guides UGV path planning through a risk-averse shared exposure belief and uses mission-conditioned rewards to drive UAVs to focus sensing on exposure-relevant uncertain regions; UAV observations update the belief online, and the UGV uses this to route around newly discovered threats. In multi-UAV scenarios, region-conditioned spatial assignment is used to reduce redundant sensing.
arXiv cs.RO ·