Event

CONSORT: Multi-Robot Cooperative Localization Planning under GPS Denial

1 reports · 1 sources

Tony Zhang and co-authors submitted a preprint proposing CONSORT, a multi-robot cooperative localization planning method for GPS-denied environments: the primary agent must reach the goal in minimum time under collision probability constraints and a terminal localization uncertainty bound. The method optimizes only the mean state trajectory, propagates covariance only when evaluating terminal constraints, converts the belief-space problem into a solvable form, and achieves discrete-to-continuous planning via a joint search over a hexagonal grid plus spline trajectory refinement. The authors report that across 50 randomized scenarios, planner performance diverges as the terminal uncertainty constraint tightens: CONSORT solves 80% of the scenarios, versus 54% for the strongest baseline, with comparable path lengths. In maritime simulations, when executed under feedback control and using an independently developed factor-graph estimator, trajectories remain collision-free and terminal uncertainty differs from its prediction by within 6%. These results are as reported by the authors and have not yet been independently verified.