A reinforcement learning method for human-robot collaboration in assembly tasks

• Published in: Robotics and computer-integrated manufacturing 2022-02, Vol.73, p.102227, Article 102227
• Main Authors: Zhang, Rong, Lv, Qibing, Li, Jie, Bao, Jinsong, Liu, Tianyuan, Liu, Shimin
• Format: Article
• Language: English
• Subjects: Adaptive decision; Algorithms; Alternators; Assembly; Collaboration; Human-robot collaboration; Industrial applications; Machine learning; Reinforcement learning; Robots; Task allocation; Task complexity
• ISSN: 0736-5845; 1879-2537
• DOI: 10.1016/j.rcim.2021.102227

•Compared with traditional assembly method, a reinforcement learning method for human-robot collaboration assembly proposed in this paper has the following advantages.•The behavior of agents with reinforcement learning ability is adopted to guide operators, and no longer need humans to make decisions.•The definition of the human fatigue function, and the actor network adds noise to deal with the dynamic characteristics of the human model.•The assembly task is expressed by vectorization, and the detailed definition makes the task allocation more accurate.•The human-robot collaborative reinforcement learning (HRC-RL) network is designed for the collaboration among different agents. The assembly process of high precision products involves a variety of delicate operations that are time-consuming and energy-intensive. Neither the human operators nor the robots can complete the tasks independently and efficiently. The human-robot collaboration to be applied in complex assembly operation would help reduce human workload and improve efficiency. However, human behavior can be unpredictable in assembly activities so that it is difficult for the robots to understand intentions of the human operations. Thus, the collaboration of humans and robots is challenging in industrial applications. In this regard, a human-robot collaborative reinforcement learning algorithm is proposed to optimize the task sequence allocation scheme in assembly processes. Finally, the effectiveness of the method is verified through experimental analysis of the virtual assembly of an alternator. The result shows that the proposed method had great potential in dynamic division of human-robot collaborative tasks.