TY - JOUR T1 - DERIVATION OF VELOCITY/DISTANCE CURVES FROM A WIRELESS INERTIAL MEASUREMENT UNIT INSTRUMENTED BOB-SKELETON SLED AND COMPARISON TO LIGHT GATE AND VIDEO DERIVED VALUES JF - British Journal of Sports Medicine JO - Br J Sports Med SP - e4 LP - e4 DO - 10.1136/bjsports-2013-093073.32 VL - 47 IS - 17 AU - M Gaffney AU - M Walsh AU - B O'Flynn AU - C O'Mathuna Y1 - 2013/11/01 UR - http://bjsm.bmj.com/content/47/17/e4.26.abstract N2 - The sprinting push-start in the Winter Olympic sport of Bob-Skeleton, constitutes 2% and 8% of the 1500 m distance and 70 s duration of each run (Official Results & Track Data, www.fibt.com, Apr 2013). This phase is critical to performance (Zanoletti et al. JSCR 2006;20(3):579–583), but the sport's small size, seasonal nature and other issues i.e. access to tracks, the challenging environment, or even restrictions by the governing body (Bob-Skeleton Rules: 10.11, 14.1 & 16.9, www.fibt.com, Apr 2013), mean no studies-with or sources-of fine-grained real-world data are available. A Wireless Inertial Measurement Unit (WIMU) system was used to instrument a Bob- Skeleton training sled to gather such data. A WIMU streamed 256Hz Accelerometer data; light-gate, video and track-shape data were also gathered. A 65 kg female athlete in her 20 s performed 11 runs with varying step count, push side and handedness. An adaptive integration method using track-shape compensation and filtered raw data, converted Acceleration to velocity/distance curves and compared it to values interpolated from combined timing data from 6.5–20.5 m after push-off. WIMU data from 7 runs was used (average of data points 486±96). The adaptive integration method results tracked timing data velocity/distance curves well. The average velocity error was 0.462±0.248 m/s (7.57±4.12%), best & worst runs were 0.176±0.143 m/s (2.97±2.63%) & 0.800±0.251 m/s (13.2±3.55%) respectively. WIMUs show promise for recording fine-grained Bob-Skeleton velocity/distance data, but results so far are not good enough for high-end users (<0.1 m/s accuracy). Improved systems with higher sampling, better sensors, in-the-field re-calibration, integrated timing, and synchronisation capabilities could likely achieve this. Acknowledgements Supported by Science Foundation Ireland via CLARITY: Centre for Sensor Web Technologies under grant 07/CE/I1147 with the assistance of University of Bath & UK Sports. ER -