Wiping a table cluttered with objects is a tricky task for robots. Most existing cleaning strategies simply avoid obstacles and only wipe the free areas, leaving dirty patches underneath and around the objects. A new preprint on arXiv (2610.19527) proposes a different approach: a dual-arm cooperative planning framework that explicitly handles scattered objects by temporarily moving them out of the way.
The framework plans wiping motions for two arms working together. Instead of treating objects as permanent barriers, it incorporates temporal object removal—relocating items during the cleaning process so the surface underneath can be wiped, then presumably returning them afterward. This allows more complete coverage than conventional detour-based methods.
The abstract does not provide implementation details, but the core idea is clear: cooperative, time-aware manipulation can improve cleaning quality. The authors contrast their strategy with standard surface-cleaning techniques that only cover free areas, highlighting the novelty of their temporal-removal approach.