Step Length Estimation Using UWB Technology: A Preliminary Evaluation

The information about the pedestrians' step length is useful for a great variety of applications. The technologies that are commonly selected for estimating the step length offer very high accuracies, such as millimeter-level accuracy in the case of optical systems or centimeter-level with ultrasound technology. However, their use is usually restricted to dedicated small indoor laboratories, which makes it difficult to collect realistic data belonging to day-to-day activities. For this reason, despite their high precision, these systems are not usually used in the field of inertial pedestrian navigation, in which it is common to have to train step length models. UWB technology offers only decimeter-level ranging accuracy but provides larger coverage areas. Therefore, it could be considered as a candidate technology for step length estimation purposes in more larger testing scenarios, which can be useful for the training of pedestrian navigation systems. In this paper, it is presented a preliminary evaluation of the accuracy of UWB technology for step length estimation purposes. Four different methods are implemented and tested: one based on the variations of the ranges to a fixed point, another based on the inter-feet ranges and two more methods based on the relative positions between footsteps. The method based on the inter-feet ranges obtained an average step length error of 9.9 cm $(\pm\mathbf{7.1}$ cm) and a median relative error of 15%, while one of methods based on the feet positions got an average error of 11.9 cm $(\pm\mathbf{6.3}$ cm), but only a median error of 5.9 cm if Line-Of-Sight (LOS) scenarios are only considered. These preliminary results confirm that errors of a few centimeters can be achieved. Now, further research is needed in order to evaluate if the combination of the accuracy and coverage provided by UWB is really useful for training pedestrian navigation systems.