Decentralized Path Planning in Lunar Robot Swarms: Local Optimization for Collision Avoidance under Constrained Perception

This paper investigates decentralized safe path planning for lunar robot systems in dynamic environments with limited perception and communication capabilities and proposes a decentralized planning-control framework based on local path replanning. For clarity, we first define system models for complex dynamic scenarios and multi-robot systems. Then, the decentralized framework is systematically developed through two complementary parts, that is, ensuring feasible path replanning under collision conditions and preventing secondary collisions caused by path tracking control. Concretely, path reachability analysis is applied to derive quantifiable replanning criteria based on discrete sampling intervals, supported by a reachable set algorithm that theoretically guarantees replanning feasibility. Path relevance constraints are extended through analysis of collision-prone waypoints, effectively eliminating the risks in continuous trajectory. To enhance the practical robustness, bounded uncertainties from tracking control and environment are taken into consideration in the actual application. On this basis, a comprehensive numerical synthesis is proposed. Finally, lunar base construction simulations validate the proposed decentralized planning-control framework, demonstrating its effectiveness in safe path planning and its scalability in large-scale multi-robot deployments. © 1965-2011 IEEE.