Machine learning-assisted self-powered intelligent sensing systems based on triboelectricity

The advancement of 5G and the Internet of Things (IoT) has ushered in an era of super-interconnected intelligence, which promises high-quality social development. Triboelectric-based sensing systems, combined with big data, artificial intelligence (AI), wireless communications, and cloud computing, among others, have gained considerable attention in self-powered sensing technologies. Machine learning (ML), as a critical component of AI, offers a practical strategy for efficiently processing multi-dimensional and multi-type data collected by triboelectric-based intelligent sensing systems (TISSs). In this review, a comprehensive and systematic summary of the latest advances in ML for TISSs from the perspective of technology implementation is innovatively presented. Then, we elaborate on the characteristics of common ML algorithms for TISSs and how ML broadens the applications of the triboelectric nanogenerator (TENG), which will provide valuable references for subsequent research. Finally, the limitations and prospects of ML for TENG are discussed for large-scale deployments, aiming to achieve higher-level intelligence in the future. These will provide significant insights to enable the advancement of self-powered sensing technology based on triboelectricity. [Display omitted] •A systematic summary of the latest advances in ML for TISSs from the perspective of technology is firstly presented.•The characteristics of common ML algorithms for TENG and how ML broadens the applications of TENG are investigated.•The limitations of data acquisition and algorithm selection in experimental conditions for TISSs are analyzed.•The prospects and emerging technology of ML for TENG are discussed for large-scale deployments.