Abstract: Due to the disturbance in human-robot collaborative disassembly (HRCD), such as scrambling disassembly tools by multiple robots and adjusting disassembly processes by humans, tools are often lacked or occupied for the current disassembly process and long waiting time is wasted if lack of flexible tool scheduling strategy, which impairs the efficiency of the HRCD process. Therefore, improving the ability of the robot for cognizing the current disassembly circumstance and recognizing the replaceable tools to generate the alternative solutions, counts for tackling the disturbance scenarios in HRCD. To this end, an affordance reasoning-based sequence planning manner for HRCD is proposed, which comprises three phases. Firstly, a knowledge graph is established with the structural data of products and process operating documents, so as to provide the operation and sequence information for HRCD tasks. Then, concerning a process that lacks essential tools, an affordance reasoning manner is proposed for the visual scenes of HRCD to offer the replaceable tools for this process. After that, an affordance reasoning-based sequence planning manner is designed to optimize the original disassembly sequence if tools have been replaced, trying to reduce the total disassembly time. Taking the HRCD of aging batteries in new energy vehicles as the experimental case to validate the performance of affordance-based detection and disassembly sequence planning. The results demonstrate that the proposed manner could provide accurate alternative solutions for tool replacement when essential tools are occupied, and could effectively reduce the time cost for the whole disassembly process, which enhances resilience and efficacy in HRCD.

Key words: human-robot collaborative disassembly, affordance reasoning, affordance detection, disassembly sequence planning, knowledge graph