CNS Core: Small: Collaborative Research: A Motion Tracking Library for Sports Analytics using Wireless Sensors

The project explores embedding sensors in balls, racquets, and on players to track metrics of interest in sports such as location and spin trajectories of balls, bat swings, and motion of players. The amateur sporting market is global, ranging from schools to clubs to local games among neighborhood friends. Bringing technologies that are otherwise professional-grade and expensive to the masses could result in wide-scale impact in sports analytics. This technology will encourage a healthier lifestyle among young individuals by making gym, exercise and sports even more exciting and helping them improve their skills. The motion tracking libraries to be developed and released publicly as a part of the project would catalyze innovations by allowing anyone to extend the libraries into creative applications. Further, the results will be published in academic venues and integrated into graduate and undergraduate classes, tutorials and workshops. Industry collaborations will be explored to enhance the outreach. Finally, if successful, the project can offer a valuable springboard to this broad application area, with a focus on making the technologies accessible to the masses.

Foundations for Internet of Things (IoT)-based sports analytics is already underway, but most of the efforts are focused towards establishing the platform - the sensing capabilities are primitive, at best. The difficult research questions, such as high-speed ball tracking, spin estimation, racquet motion analysis, and human-gesture decomposition, remain unaddressed. Unfortunately, the rich literature in wireless localization and inertial gesture recognition do not apply under the constraints for real sporting environments. For instance, Wi-Fi based localization is not designed to support centimeter-scale 3D-location at ball speeds. Inertial sensors such as accelerometers cannot measure gravitational direction, whereas gyroscopes saturate under ball rotations. This proposal will cut across and help bring together many research areas like motion sensing, wireless networking, statistical inference, circuit design, kinesiology, and sports. The proposal will enable building sports analytics systems from cheap wireless IoT sensors which can replace the very expensive vision-based systems. To this end, it will it will focus on fusion of under constrained sensor data with aerodynamic models of ball motion, patterns of bat swings, kinematic models of arm and body motions to track location, spin, and motion of balls, bats, and players. Statistical inferencing techniques would be used to estimate the parameters under noisy sensory information. Finally, an app will be developed to visualize the motion and provide feedback to players.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.