Dual-Arm Tabletop Rearrangement with Movable Stacking and Gap-Aware Push Transfers

This research presents a dual-arm rearrangement framework that integrates movable stacking with gap-aware push transfers to facilitate asynchronous cross-boundary object exchanges in cluttered tabletop environments. The method identifies collision-free transfer gaps along shared boundaries, places objects near selected crossing points, and pushes them into receiving workspaces without requiring simultaneous arm interaction. A joint Monte Carlo Tree Search (MCTS)-based planner coordinates both manipulators while reasoning over buffering and transfer decisions, with execution cost used post-planning to assess plan quality. Experiments demonstrate that the proposed approach shortens joint execution sequences and outperforms fixed handoff baselines as scene density increases, achieving lower execution costs in denser scenes while remaining competitive in lighter ones.