Optimal Task Allocation in Human-Robotic Assembly Processes

Abstract

This study aims to design a semi-automatic assembly line that is relevant to human-robot task allocation problems. It combines two methods, which are Design for Assembly (DFA) and optimization. First, the DFA difficulty score of each task including inspection is evaluated when it is performed by humans and robots. The score is then put in an optimization model. A mathematical model optimally assigns tasks to humans and robots with a feasible sequence. The proposed mathematical models are illustrated on a Lego-car assembly with two demand scenarios, being low and high. Results show that while three single objective models do not provide good solutions, a multi-objective linear problem (MOLP) minimizing a total cost, a cycle time, and difficulty scores altogether provides a better solution. The weights of objectives in MOLP are determined by a modified two-person zero-sum game with a weighted sum method.