Smart triboelectric motion-feedback sensor for table tennis monitoring

With the advancement of flexible electronics and machine learning (ML), the collection and analysis of athletic big data using self-powered sensors have become increasingly important in intelligent sports. However, accurate quantification of the rotation direction, a critical parameter in ball games, remains largely underexplored. This study proposes an ML-assisted, intelligent motion-feedback table tennis paddle integrated with a triboelectric nanogenerator (TENG) for simultaneous velocity and rotation direction sensing. By leveraging a well-intercalated MXene/bacterial cellulose (MXBC) composite thin film, the developed sensor exhibits improved tensile strength, excellent conductivity, good processability, and lightweight characteristics. Consequently, the MXBC-based TENG generates distinct and rich feature signals corresponding to rotational motions, enabling its use as a self-powered intelligent feedback sensor. To enhance detection sensitivity and accuracy, a hybrid long short-term memory network–support vector machine (LSTM–SVM) model is constructed, coupled with effective denoising technique. Using the optimized LSTM–SVM modal, the proposed sensor achieves table tennis ball rotation direction and velocity recognition with an accuracy of 99.8%, representing a 48.5% improvement over the conventional SVM method. This study is the first to achieve self-powered rotation direction detection in table tennis, demonstrating significant potential for ball sports training and opening new avenues for the advancement of intelligent sports.