An improved fault-tolerant cultural-PSO with probability for multi-AGV path planning

This paper presents a hybrid evolutionary algorithm, cultural-particle swarm optimization (C-PSO), which is inspired by the cultural algorithm and the particle swarm optimization algorithm. It is aimed to balance the performance of exploration and exploitation and avoid trapping in the local optima. It introduces a probabilistic approach to update the inertia weight based on the improved metropolis rule. Generating the optimal path without collisions is challenging to ensure vehicles operate safely in real-time implementation. The contributions of C-PSO are to solve the path planning problem of multiple vehicles in modern industrial warehouses, achieving task allocation, fault tolerance and collision avoidance by a dual-layer framework. It was compared with the other algorithms, including PSO, PSO-GA, CA, HS, ABC, HPSGWO, TS and MA, by CEC benchmark functions and statistical tests to demonstrate its great performance with fewer iterations and runtime and the best solutions. It is validated through computational experiments, which involve 15 AGVs and 20 tasks for demonstration.