Event
SkyAgent improves aerial search with revisable target memory
1 reports · 1 sources
Kaibo Yang et al. propose Revisable Target Memory (RTM) in SkyAgent to address the problem in persistent aerial search where old candidates still trigger approach or verification actions after contradictory evidence (which the authors call false candidate persistence, FCP). RTM associates evidence with persistent candidates, suppresses repeated counting of correlated viewpoints, and allows candidates to be revised among three states: Tentative, Confirmed, and Rejected. The authors report that on four splits across three benchmarks—CityAVOS, UAV-ON, and OpenUAV—SkyAgent equipped with RTM outperforms all compared methods, improving success rate by 2.90–6.31 points and markedly reducing Behavior-FCP and movement after contradictions, with a low false-rejection rate; in real DJI scenarios with 35 matches per method, RTM also improved task success rate and behavioral reliability.
Report timeline
- SkyAgent proposes Revisable Target Memory (RTM) to improve continuous aerial search
SkyAgent proposes Revisable Target Memory (RTM) to address the problem in continuous aerial search where, after evidence becomes contradictory, old candidates still trigger approach or verification actions (which the authors call false candidate persistence, FCP). RTM associates evidence with persistent candidates, suppresses repeated counting of correlated views, and lets candidates be revised among three states: Tentative, Confirmed, and Rejected.
arXiv cs.RO ·