Three machine learning techniques for automatic determination of rules to control locomotion

Automatic prediction of gait events (e.g., heel contact, flat foot, initiat ion of the swing, etc.) and corresponding profiles of the activations of mu scles is important for realtime control of locomotion. This paper presents three supervised machine learning (ML) techniques for prediction of the act ivation patterns of muscles and sensory data, based on the history of senso ry data, for walking assisted by a functional electrical stimulation (FES), Those ML's are: 1) a multilayer perceptron with Levenberg-Marquardt modifi cation of backpropagation learning algorithm; 2) an adaptive-network-based fuzzy inference system (ANFIS); and 3) a combination of an entropy minimiza tion type of inductive learning (IL) technique and a radial basis function (RBF) type of artificial neural network with orthogonal least squares learn ing algorithm. Here we show the prediction of the activation of the knee fl exor muscles and the knee joint angle for seven consecutive strides based o n the history of the knee joint angle and the ground reaction forces, The d ata used for training and testing of ML's was obtained from a simulation of walking assisted with an FES system [39], The ability of generating rules for an FES controller was selected as the most important criterion when com paring the ML's. Other criteria such as generalization of results, computat ional complexity, and learning rate were also considered, The minimal numbe r of rules and the most explicit and comprehensible rules were obtained by ANFIS. The best generalization was obtained by the IL and RBF network.