This paper presents a decentralized approach for collision-free navigation in multi-robot systems, where each robot independently computes its own path while coordinating locally with neighboring robots to avoid collisions. Unlike centralized path planning methods, the proposed approach eliminates reliance on a global controller and instead emphasizes robot-level autonomy and local interaction. Each robot initially plans a path using a classical single-agent planner and dynamically adjusts its motion when potential collisions are detected. Local coordination strategies such as temporary waiting, speed adjustment, and resolve conflicts in real time. To validate this proposed framework, we developed a comprehensive multirobot simulation environment using Python. The system was tested on various grid maps with static obstacles to evaluate performance metrics such as path efficiency and collision counts. The approach improves scalability, robustness, and adaptability in dynamic environments. Simulation-based experiments demonstrate that the proposed method achieves zero collision rates while maintaining acceptable path efficiency and computational performance. The results indicate that independent path computation combined with local coordination is a practical and effective solution for decentralized multi-robot navigation.
Hatwar et al. (Mon,) studied this question.