A Motion Classification Algorithm of Lower-Limb Rehabilitation Robot

When patients with lower limb motor dysfunction do active rehabilitation training, the rehabilitation robot has to predict the patient’s posture accurately. A human motion posture recognition algorithm based on a Naive Bayes classifier is proposed for improving the motion recognition accuracy of the rehabilitation robot. Firstly, a multi-sensor information acquisition system is built on the rehabilitation robot to collect the shoulder displacement data and the forces magnitude on both sides of sagittal plane during the rehabilitation process of the patient. Then, the information acquisition system sends the acquisition data to the upper computer through wireless communication for generating the training data set. Finally, the data set is processed by Naive Bayes classifier to calculate the current human action and predict the next moment action. Simulation and experimental analysis of the Naive Bayes algorithm show that the classifier can constantly update the posterior probability based on the conditional probability and achieve a high recognition rate.