Evaluating Multi-Agent Path Finding Algorithms for Modular Robot Reconfiguration

Modular self-reconfigurable robots are composed of many connected modules that must move in coordination to transform from one shape to another. Effective reconfiguration requires Multi-Agent Path Finding (MAPF) methods that manage both spatial and temporal conflicts; in this paper, we compare three such coordination strategies: Cooperative A* (CA*), Conflict-Based Search (CBS), and a flow-based pairing and streamline construction method. Our results show that CA* provides fast and lightweight planning suitable for sequential or limited-parallel motion. CBS produces lower-cost solutions and stronger conflict handling but requires more computation as the number of conflicts increases. The MaxFlow approach offers an efficient way to assign targets and organize movement, reducing ambiguity and improving coordination during structured reconfiguration stages.