AI-Embedded Motion Sensors for Sports Performance Analytics

Motion sensing technology is widely used in healthcare, sports, consumer electronics, etc. On the other hand, Artificial intelligence (AI) enables the development of wearable sensors that can recognize and analyze human motions. In this study, we developed a wearable wireless motion sensing system with AI-embedded inertial measurement units (AIMUs), combined with visual analysis, to evaluate and optimize athletes’ performance in sports such as gymnastics. In the experiment, a gymnast performed an entire vaulting routine while wearing 11 AIMUs, and the motion data were transmitted to a cloud server through Bluetooth gateways. This system can achieve the segmentation of vaulting phases and the evaluation of detailed movements. The experimental results showed a 4.57% estimation error in flight height. This AI-embedded motion sensor system has the potential to provide an intuitive and easy-to-understand way to present athlete performance to coaches and athletes themselves. Furthermore, its continued development could assist athletes’ training, provide quantitative sports performance indicators, significantly improve elite athletes’ training efficiency, and monitor their health regularly.