RoboLeader: Dynamic Re-Tasking for Persistence Surveillance in an Urban Environment Using Robot-to-Robot Control

摘要

Abstract : In the FY09 DRI, we developed the RoboLeader intelligent agent that had the capabilities of coordinating a team of ground robots and revising route plans for the robots based on battlefield intelligence. In the current DRI, the capabilities of RoboLeader were expanded to deal more specifically with dynamic re-tasking requirements based on battlefield developments as well as coordination between aerial and ground robots in pursuit of moving targets. The results of our human-in-the-loop simulation experiment showed that RoboLeader (Fully Automated condition) was more effective in encapsulating the moving targets than were the human operators (when they were either without assistance from RoboLeader [Manual] or when they were partially assisted by RoboLeader [Semi-Autonomous]). Participants successfully encapsulated the moving targets only 63% of the time in the Manual condition but 89% of the time when they were assisted by RoboLeader. Those participants who play video games frequently demonstrated significantly better encapsulation performance than did infrequent gamers; they also had better SA of mission environment. Visualization had little effect on participants? performance. Finally, participants reported significantly higher workload when they were in the Manual condition than when they were assisted by RoboLeader.