Edge-Based Real-Time Analytics for Mobile Sports Performance Tracking

Mobile edge computing, IoT, and AI are innovative concepts in the fields of monitoring the mobile sports performance, with high hazards in comparison to the previous time regarding latency and data quality of the real-time analysis. However, all the existing solutions either are limited to the input of a single-sensor or limited to standalone metrics without the integration of different data streams that would generate the complete and functioning intelligence. Within this direction towards eliminating this dire gap, we introduce EdgeTrack, a new multi-modal analytics pipeline, entirely executed at the edge. The EdgeTrack is a combination of physiological (heart rate, SpO 2, etc.) and biomechanical (motion, acceleration, posture, etc.) sensor information and gives real time, contextually adaptive responses to athletes and coaches. It relies on a sensor fusion on-device, lightweight context-aware AI models, and resource allocation that can offer low-latency monitoring with high precision of various variables across various domains of sports. EdgeTrack, which can be used alongside the current edge System-on-Chips (SoCs) commercially, it does not have to connect to cloud computing infrastructure, it has a significantly faster response time without affecting the privacy of the user and wasting power. The platform of EdgeTrack has demonstrated itself in the practical environment of interactions with athletes of different sports to be more superior than the older system of cloud based and single modality. It realizes 60 percent, 12 percent more accurate and 35 percent lower energy consumption results. Although this is done individually, EdgeTrack will enable athletes to accept in-the-moment performance decisions and real-time data and coaches to have a better picture of the situation. It is a recent standard of intelligent sport tracking and the application is very extensive in team training, injury recovery and recreational exercise.

Hali tarjima qilinmagan