In long distance running, real time monitoring and performance optimization has been recently rendered possible through the commercialization of a large variety of running Computers. Generally, such Computers simultaneously monitor several parameters such as heart rate, running speed, stride frequency and stride length. More precisely, stride and speed information is obtained using foot located inertial sensors. Unfortunately, due to its leg extremity location, the foot inertial sensor presents some disadvantages: it may be perceived as cumbersome; it requires supplementary telecommunication facilities as well as local power supply; it may increase a runner's energy expenditure even though it weights only a few tens of grams. To overcome the above-mentioned drawbacks we have developed a method for the estimation of a runners speed, stride frequency and stride length based on inertial sensors which may directly be integrated in classical chest-belt heart-rate-monitors. The method processes various kinetic features from chest located accelerometers. The instantaneous speed is then estimated through a linear mapping of the most reliable kinetic features. The determination of the most reliable kinetic features as well as the mapping coefficient is achieved through a minimal variance approach on a calibration run. A validation has been performed on 9 subjects on various running surfaces and over durations of l year in order to highlight the consistency of the proposed approach. We have found that different runners necessitate different optimal weighting factors of the kinetic features. This may be related to a runner's stride efficiency, that is, more or less energy wasting and bouncing trajectories for similar forward speed. However, for a given runner we obtained consistent results namely a speed estimation accuracy of about 7% in a range of 60% to 110% of his maximal aerobic speed.
© Copyright 2008 The Engineering of sport 7, Volume 1. Veröffentlicht von Springer. Alle Rechte vorbehalten.