Path Planning for 4-Wheeled Omnidirectional Cellular Conveyor using Q-Learning Algorithm

The manufacturing industry has implemented various technologies, such as robot arms, AGVs (Automated Guided Vehicles), and conveyors, to enhance the efficiency of material flow. Conveyors facilitate the transportation of objects over long distances. The fixed belt conveyor is the most commonly used type of conveyor across various industries. The fixed belt conveyor follows a predetermined path to distribute items. Nevertheless, during its development, the material distribution process requires various paths. Therefore, the development of flexible conveyors was necessary to meet the requirements of various delivery routes. The most recent innovation in flexible conveyors is the conveyor with omnidirectional wheel drive. This study involved developing a path-planning algorithm utilizing classic Q-Learning (CQL), Double Q-Learning(DQL), and Action Restrictions Q-Learning (ARQL) to calculate the optimal path on an four-wheeled omnidirectional conveyor. The results show that the ARQL algorithm generates the optimal trajectory, achieving a processing time of less than 100 milliseconds.