Author(s): CUI, Y., YANG, W., WANG, J., Institution: BEIJING SPORT UNIVERSITY (BSU), Country: CHINA, Abstract-ID: 2200

INTRODUCTION:
The application of player tracking systems automatically provide time-motion and tactical information for athletic performance analysis purposes. Compared to the GNSS based system with lower accuracy and optical tracking system with higher cost, the local positioning system seems to be feasible for racquet sports of smaller court area. The aim of this study was to comprehensively verify the accuracy of a ultra-wideband (UWB) player tracking system for monitoring tennis-specific movements.
METHODS:
Ten amateur tennis players (International Tennis Number: 2 to 5) were recruited in the study and wore the UWB device (GenGee Insait KS) on their left upper arm during the whole experiment. An infrared camera-based motion capture system with 16 cameras (VICON, Oxford, UK) was employed as the reference to determine criterion position. Five adhesive retro-reflective markers with a diameter of 10mm were fixed to each participant’s skin to determine the center of mass. All participants performed the following 4 exercises without racquets and 3 with racquets: warm-up (walk, jog, acceleration, side-shuffle), agility ladder, spider run, T-run, hit-and-turn, fixed tactics, free rallying. Each drill was repeated at least twice expect for the last one, with adequate rest time allocated for participants. The raw derived data from two systems were processed to calculate the distance traversed. The linear mixed model, intra-class correlation coefficient (ICC) and root mean square error (RMSE) were used to estimate the accuracy of the UWB-based system.
RESULTS:
The smallest differences of RMSE in raw x and y coordinate accuracy were shown for agility ladder (0.31m) and walk (0.40m), while largest RMSE in x and y were shown for spider run (0.90m) and T-test (1.07m). The ICC statistic of 0.913 was showed for the distance covered during the while experiment measured by UWB system with that by the reference system, indicating almost perfect agreement. However, noticeable exercise-dependent variation in measurement accuracy were exhibited, particularly in agility ladder, T-run and fixed tactics (ICC: -0.28—0.14), which were also confirmed by the significant difference from the linear mixed model (p<0.05).
CONCLUSION:
The UWB-based device presented acceptable accuracy in recording tennis-specific movement, but noticeable difference from the criterion was shown exercises with high intensity acceleration, deceleration and change of direction, suggesting large error margin of measurement. Apart from the inherent error, the arm-fixed location of UWB device might also led to the discrepancy in measuring actual distance in these drills, as arm swing and early hip rotation are inevitable. The application of the system should be done with caution given its potential overestimation of external load, and practitioners are advised not to directly compare the data with other tracking system during training and competition scenarios.