Reinforcement Learning And The Application To Intelligent Warehousing

Intelligent warehouses are an essential part of the material-handling industry. Efficient and sensitive warehousing with intelligent mobile robots is critical to improve the overall productivity and simultaneously achieve a high efficiency. This paper focuses on reinforcement learning and its application to intelligent warehousing. Reinforcement learning is a powerful mechanism that does not require any prior knowledge and the robot can autonomously learns to improve its performance overtime by the trial-and-error interacting with its working environment.A new hybrid solution is presented to improve the efficiency of intelligent warehouses with multi-robot systems, where the genetic algorithm based task scheduling is combined with reinforcement learning based path-planning for mobile robots. Reinforcement learning is an effective approach to search for a collision-free path in unknown dynamic environments. Genetic algorithm is a simple but splendid evolutionary search method that provides very good solutions for task allocation. In order to achieve higher efficiency of the intelligent warehouse system, we design a new solution by combining these two techniques and provide an effective and alternative way compared with other state-of-the-art methods. Simulation results demonstrate the effectiveness of the proposed approach regarding the optimization of travel time and overall efficiency of the intelligent warehouse system.In a multi-agent reinforcement learning system, the action selection of an exploring agent may be affected by other agents’actions. In other words, joint-states and joint-actions should be studied. The circumstance of a multi-agent system is dynamic and complex. In order to reduce the computational complexity and improve the learning efficiency, CQ-learning which based on sparse interactions is proposed to solve the multi-robot path planning problem. Furthermore, transfer learning has been introduced during the process of learning. Based on the prior knowledge about the environment, the learning agent can learn much faster. The experimental results demonstrate the effectiveness of the new approach.