Safe Multistage Reconfiguration Motion Control for Self-Reconfigurable Swarm.

Modular self-reconfigurable robot (MSRR) swarms offer significant potential for structural adaptation in unstructured environments. However, achieving efficient and precise reconfiguration is challenging, as the process must transform scattered individuals into cohesive assemblies while satisfying decentralized coordination, nonholonomic motion constraints, and strict terminal pose requirements. To address these challenges, this work presents a systematic three-stage distributed framework for autonomous reconfiguration. First, feasible assembly poses are generated and assigned through a decentralized negotiation mechanism, enabling the swarm to determine target configurations without a central controller. Second, a safety-aware multirobot trajectory planning method is formulated to generate collision-free trajectories under nonholonomic kinematic constraints and terminal position-orientation constraints. Finally, a vision-based alignment strategy is designed to ensure high-precision docking at the terminal stage. Simulation and real-world experimental results demonstrate the practical feasibility, favorable scalability, and reliable reconfiguration performance of the proposed framework under the tested conditions.