A human-AI agents collaborative disassembly depth reasoning method for end-of-life automotive batteries under uncertain operating conditions

End-of-life automotive batteries often exhibit uncertain operating conditions typical of a VUCA environment, including structural deformation, component corrosion, and cell performance degradation, making it difficult for human operators to determine an appropriate disassembly depth based solely on experience or fixed rules. To address this problem, this paper proposes a Human-AI Agents collaborative disassembly depth reasoning method for end-of-life automotive batteries under uncertain operating conditions. First, a few-shot battery state perception and anomaly detection method is developed to identify key components and detect unstructured anomalous states in complex human-robot collaborative disassembly scenarios. Second, a multidimensional feature quantification and fuzzy evaluation method is constructed to transform anomaly detection results into structured component state profiles for subsequent reasoning. Third, a Human-AI Agents collaborative disassembly depth reasoning framework, named B-MADD, is proposed to collaboratively integrate cell-state parameters, the component anomaly state profile, historical cases, and economic evaluation results, enabling adaptive and interpretable disassembly depth recommendations under uncertain operating conditions. A case study and experiments were conducted using an end-of-life battery disassembly dataset collected from real human-robot collaborative disassembly scenarios. The results show that the proposed perception module achieves a mean class accuracy of 85.7% in key battery component recognition, while the reasoning framework reduces the average deviation between the predicted and optimal disassembly depths to 0.20. These results demonstrate the effectiveness of the proposed framework and provide methodological support for intelligent and human-centered disassembly decision-making, contributing to safer, more efficient, and more sustainable end-of-life battery recycling.