Multi-robot Path Planning Based on Proximal Policy Optimization Algorithm

The ‘goods to people’ model-based intelligent warehouse employs an automated dispatching system, primarily reliant on handling robots, thereby significantly reducing manpower and enhancing the efficiency of warehouse operations. This study tackles a paramount challenge in intelligent warehouse systems, specifically, multi-robot path planning for express package handling in the intelligent warehouse by comprehensively considering different movement speeds of robots and multiple target locations (package unloading locations) need to be arrived in sequence for each robot. To address the multi-robot path planning problem, we utilize the Proximal Policy Optimization (PPO) algorithm integrated with centralized, decentralized, and hybrid policy optimization networks. The simulation experiment indicates that the implemented PPO, in conjunction with a hybrid policy optimization network, outperforms in orchestrating multi-robot path planning.