Industrial application of the Bipartite TSP in robotic pick-and-place operations

This paper addresses the challenge of optimizing pick-and-place robotic operations using a Kitting Robot System in the pre-assembly stage of car manufacturing. The objective is to minimize the time required to transfer car components from containers to kit-holders for subsequent assembly. We model this problem as a Bipartite Traveling Salesman Problem (BTSP) and apply various solution approaches. Specifically, we utilize two well-known heuristic methods, namely Nearest Neighbor (NN) and the 2-opt approximation algorithm, as well as an Integer Programming model adapted for BTSP. Moreover, we explore the applicability of a Reinforcement Learning approach based on Q-learning. By analyzing computational times and solution quality, we provide insights into the efficiency of these methods in an industrial setting, highlighting their suitability based on problem size and complexity.