A New Incremental Search Method for Multi-robot Path Planning

Multi-robot path planning problem involves some challenges. One of them is the exponential increase in the size of the search space as a result of increasing the number of robots in the operating environment. Therefore, there is a need for algorithms with high computational performance to plan optimal and collision-free paths in a limited time. In this article, a centralized path planning algorithm is presented. The algorithm is a heuristic incremental search, in which the D* Lite algorithm has been adapted for the multi-robot case. The concept of occupancy time has been embedded into the environment model to avoid path interference. A centralized function has been designed to update the environment model and robot data. To evaluate the method, two groups of simulations of static and dynamic types were carried out. The static simulations focused on studying the effect of algorithm parameters, and it was shown that the algorithm can plan paths for up to 40 robots in an environment having 55 percent free space. The dynamic simulations were carried out in Gazebo, a real-time and dynamic physical simulator. The results were compared to a baseline method based on potential fields. The number of robots was increased to 14, and it was demonstrated that for 9 robots and more, the potential field approach either fails or has a rapid increase in computation time, while the proposed method can find feasible solutions in a limited time.