ADAPTIVE REACTIVE JOB-SHOP SCHEDULING WITH REINFORCEMENT LEARNING AGENTS

Traditional approaches to solving job-shop scheduling problems assume full knowledge of the problem and search for a centralized solution for a single problem instance. Finding optimal solutions, however, requires an enormous computational effort, which becomes critical for large problem instance sizes and, in particular, in situati ons where frequent changes in the environment occur. In this article, we adopt an alternative view on production scheduling problems by modelling them as multi-agent reinforcement learning problems. In fact, we interpret jobshop scheduling problems as sequential decision processes and attach to each resource an adaptive agent that makes its job dispatching decisions independently of the other agents and improves its dispatching behavior by trial and error employing a reinforcement learning algorithm. The utilization of concurrently and independently learning ag ents requires special care in the design of the reinforcement learning alg orithm to be applied. Therefore, we develop a novel multi-agent learning algorithm, that combines data-effi cient batch-mode reinforcement learning, neural network-based value function approximation, and the use of an optimistic inter-agent coordination scheme. The evaluation of our learning framework focuses on numerous established Operations Research benchmark problems and shows that our approach can very well compete with alternative solution methods.