Review on Global Path Planning Algorithms for Autonomous Navigation of Multiple Mobile Robots

Global path planning enables efficient and collisionfree navigation in multi-robot systems operating in dynamic environments. This paper comprehensively reviews global path planning algorithms, categorizing their applicability in known and unknown environments. Classical algorithms such as $A^{*}$, Dijkstra’s, and RRT are analyzed alongside advanced methods, including Genetic Algorithms, Deep Q-Learning, and hybrid approaches like Improved A*-DWA and ACO-DWA. The study highlights critical trade-offs between computational efficiency, path optimality, and adaptability, particularly in scenarios requiring real-time obstacle avoidance and scalability. In structured settings, centralized systems demonstrate synchronized path allocation but face scalability and network dependency limitations. In contrast, decentralized approaches enhance robustness through independent decision-making at the cost of coordination complexity. Sensor integration (LiDAR, SLAM) and real-time mapping emerge as essential for minimizing collisions in known terrains, while adaptive algorithms such as RRT* and deep reinforcement learning prove effective in unknown environments requiring continuous exploration and dynamic adjustments. The review underscores the necessity for hybrid frameworks that balance exploration-exploitation trade-offs and address challenges like computational overhead and sensor reliability.