Solving Bilevel Multi-Robot Cooperative Path Planning Problems via a Memetic Framework

With the increasing use of multi-robot systems in emergency scenarios, collaborative path planning for robots has attracted greater attention. The multi-robot path-planning problem was modeled as a bilevel cooperative path planning model and solved using a memetic algorithm with a dynamic window approach and a parking scheduling strategy (MA-DWAPSS). The bilevel path planning model has divided the problem into two parts: global (static) path planning to find a near-optimal route and dynamic path planning to avoid path conflicts. Corresponding to the proposed MA-DWAPSS method, an improved memetic algorithm was developed based on genetic algorithm to find an optimal global path and a cubic Bézier curve to smooth the path and avoid sharp turns. The dynamic window approach (DWA) and parking scheduling strategy (PSS) obtain real-time sensor data and coordinate the docking and movement of robots in dynamic environments, handling obstacles in real time and preventing conflicts or unnecessary stops to improve efficiency. DWA further accounts for the dynamic characteristics of robot motion, making the path planning flexible and adaptive to rapid environmental changes. Simulation results show that the proposed method outperforms three other algorithms in path distance, time, obstacle avoidance, and smoothness.