Estimation of speed and incline of walking using neural network

A portable data logger is designed to record body accelerations during human walking. Five subjects walk first on a treadmill at various speeds on the level, and at positive and negative inclines. Then, the subjects performed a self-pace walking on an outdoor test circuit involving roads of various inclines. The recorded signals are parameterized, and the pattern of walking at each gait cycle is found. These patterns are presented to two neural networks which estimate the incline and the speed of walking. The results show a good estimation of the incline and the speed for all of the subjects. The correlation between predicted and actual inclines is r=0.98, and the maximum of speed-predicted error is 16%. To the best of our knowledge these results constitute the first speed and incline estimation of level and slope-unconstrained walking. >