Monte Carlo Tree Search for online decision making in smart industrial production

Abstract

This paper considers the issue of rapid automated decision making in changing factory environments, situations including human-robot collaboration, mass customisation and the need to rapidly adapt activities to new conditions. The approach taken is to adapt the Monte Carlo Tree Search (MCTS) algorithm to provide online choices for the possible actions of machines and workers, interleaving them dynamically in response to the changing conditions of the production process. This paper describes how the MCTS algorithm has been adapted for use in production environments and then the proposed method is illustrated by two examples of the system in use, one simulated and one in a physical test cell.