Long-Horizon Multi-Robot Rearrangement Planning for Construction Assembly

Robotic assembly planning enables architects to explicitly account for the process during design phase, and efficient building methods that profit from robots' different capabilities. Previous work has addressed of robot sequences identifying feasibility architectural designs. This paper extends previous by enabling with large, heterogeneous teams robots. We present a system which parallelization complex task motion problems iteratively solving smaller subproblems. Combining optimization solve manipulation constraints sampling-based bi-directional space-time path planner us plan cooperative multi-robot unknown arrival-times. Thus, our solver allows completing subproblems tasks differing timescales synchronizes them effectively. demonstrate approach on multiple case-studies show robustness over long horizons scalability many objects agents algorithm. Finally, we also execution computed plans two arms showcase in real world.