A gait recognition system based on BP neural network and plantar pressure
Tengda Shi, Kaifeng Su, Linqian He, Lei Yan
School of Technology, Beijing Forestry University, Beijing, China, 100083
In order to get a faster, more effective and stable control over lower extremity exoskeleton of power assist robot, precise examination on gait information is necessary, thus it is so important to design and establish a gait recognition system with accurate detection. In this paper, a wireless in-shoe wearable plantar pressure acquisition system based on ATmega16 and 8 FSR sensors will be applied to data acquisition for the gaits which consist of standing, walking, jumping and going upstairs. And four volunteers (2 males and 2 females) will be invited in this research to collect the pressure information. The NNT of MATLAB will be applied to establish an 8-12-4 BP neural net model. The input factors come from the eight sensors of plantar pressure system, the output is gait category. Proved by a great deal of experiments, the gait recognition method proposed in this research is quite feasible.