Key points are not available for this paper at this time.
Robots are becoming an increasingly important part of modern industrial production and human life. Therefore, reliable fault diagnosis is a critical task for protecting robotic systems from unexpected failures that can threaten safety and cause major economic losses. This review includes recently proposed artificial intelligence (AI)‐based techniques for robot fault diagnosis. It discusses efficient data fusion methods and categorizes multimodal information sources, including sensor‐acquired signals, internal control signals, and other state information. The discussion then moves to end‐to‐end deep diagnosis algorithms for robots. Data sparsity is addressed through transfer learning and physics‐informed paradigms. The roles of embodied AI are outlined in full lifecycle robot health management and in improving safety during human–robot collaboration. Explainable AI and model lightweighting for real‐time edge deployment are identified as promising directions for future research.
Qiao et al. (Thu,) studied this question.