Heuristic-Based Task Assignment and Multi-Agent Path Planning for Automatic Warehouses

In recent years, the problem of task assignment and path planning for multiple agents in automated warehouses, known as Task Assignment and Muti-agent Path Finding (TAPF) problem, has gained increasing attention. The TAPF problem aims to find task assignments and conflict-free path plans for agents while optimizing the objective function. However, existing algorithms are not well-suited to solve the TAPF problem in large-scale and densely populated environments while ensuring optimality. In this paper, we propose the Add Heuristics to Conflict-Based Search with Task Assignment (CBSH-TA) algorithm as an extension of the CBS-TA algorithm. Our algorithm introduces several key enhancements: 1) classifying conflict types and prioritizing conflicts, 2) incorporating heuristics into task assignment to expedite the search for optimal assignments, and 3) integrating CG, DG, and WDG heuristics into the high-level search process. We compare the proposed algorithm with the CBS-TA algorithm on maps of different sizes and densities. Experimental results demonstrate that CBSH - TA significantly reduces the number of expanded nodes, decreases runtime, and requires less memory, making it suitable for large-scale and densely populated environments.