Human motion intent learning based motion assistance control for a wearable exoskeleton

摘要: Abstract Human motion intent (HMI) acquiring by using physical human robot interaction (pHRI) information is one of the most crucial issues for lower extremity exoskeleton control. The mapping from pHRI to HMI complicated and nonlinear since wearer in control loop, which difficult be modeled directly via mathematical tools. approximation can learned machine learning approaches, e.g., Gaussian Process (GP) regression, suitable high-dimensional small-sample regression problems. However, GP restrictive large scale datasets due its computation complexity. In this paper, an online sparse algorithm proposed learn HMI, where input signal output angular increment active joints, i.e., knee joints. data HRI collected torque sensor position joint measured optical respectively. dealt with Kalman smoother achieve following functions, (1) eliminating noise (2) predicting forward. regarded as reference trajectory exoskeleton. A fuzzy-PID strategy designed drive robotic follow estimated HMI. Prototype experiments are performed on subjects who wear system walk different terrains without any transition. experimental results validated effectiveness algorithm. capable based shadow quite well.