Research on CBS Algorithm Based on Dynamic Time Windows

Multi-agent path planning in complex environments faces issues such as path conflicts and deadlocks, and the traditional CBS algorithm with a fixed time window is inefficient in handling dynamic environments. To address this, this paper proposes a Conflict-Based Search algorithm with dynamic time windows (DTW-CBS), which introduces a potential function to dynamically adjust the time window size based on the real-time progress of agents and conflict risks. This improves path planning efficiency, reduces conflicts, optimizes path quality, and enhances system throughput. Experimental results show that DTW-CBS outperforms the traditional CBS algorithm in execution time, path quality, and system throughput, especially in high-density and complex environments. However, the performance of DTW-CBS in extremely high-density environments still has room for improvement.