Abstract
An algorithm for ambulatory measurement system of the human walking is discussed. Based on the variation of the acceleration waveforms at the waist, the proposed algorithm identifies the sort of the gaits, e. g. gaits while walking on a flat floor or going up/down stairs. The procedure of the studied method is as follows: First, several acceleration waveforms for each sort of the gaits are measured, and to get a set of reference waveforms, they are classified using the clustering method. Next, unknown waveforms are compared with the gotten references and are identified. In this paper, we discuss the details of the algorithm and show the experimental results which indicate that the method successfully identifies the sort of the gaits.
