Triboelectric Nanogenerators for Human‐Robot Motion Energy Harvesting and Sensing: Structures, Embodied Artificial Intelligence, and Self‐Powered Applications

The rapid development of humanoid robots and embodied artificial intelligence (EAI) is accelerating the transformation of elderly care and rehabilitation toward interactive, adaptive, and personalized human-robot collaboration. In this context, motion sensing technology is of critical importance. Triboelectric nanogenerators (TENGs), which convert biomechanical motions into electrical signals, offer a promising route toward self-powered motion perception in embodied healthcare. This review summarizes recent advances in TENG-based multimodal motion sensing for human-robot interaction, covering inertial vibration sensing, tactile sensing, non-contact sensing, continuous trajectory sensing, and joint rotation sensing. The influences of structural configuration, motion mode, sensing direction, detection range, sensitivity, resolution, electrical output, and channel number on sensing performance under specific application scenarios are analyzed. Particular attention is paid to how TENG design strategies influence sensing universality, signal effectiveness, and the reliability of health monitoring and rehabilitation evaluation. Furthermore, the integration of TENGs with artificial intelligence (AI) technologies, including multimodal fusion, network architectures, edge computing, transfer learning, and large language models, is reviewed. Finally, applications in adaptive rehabilitation sensing, self-powered electrical stimulation, fully self-powered systems, and human‑robot interfaces are highlighted, offering perspectives on closed-loop "sensing-evaluation-regulation" frameworks for green, low-carbon EAI healthcare.