Machine Learning Methods for (Dis-)Assembly Sequence Planning - A Systematic Literature Review

This paper presents a systematic literature review on the application of reinforcement learning in the domain of assembly and disassembly sequence planning. The authors conduct a keyword search to identify scientific publications in the desired field in three scientific databases. Web of Science, Scopus and IEEE-Xplore. The analysis covers two core aspects of reinforcement learning, namely the definition of the reward function and the representation of states. In total 23 publications are identified, and the content of the collected works is presented. An analysis of the selected publications is then carried out in relation to the questions posed in order to be able to make recommendations for the application of reinforcement learning methods for the generation of efficient assembly and demonstration sequences.