Design of User motion intention algorithm for improving accuracy for Active upper limb exercise Robot

Abstract

In this paper, an algorithm for recognizing the user’s intention by utilizing single channel electromyogram sensor and force sensor is proposed as a method for generating a synchronization signal of an active-type upper limb motion robot. In active-type upper limb motion robot system, a physical human-robot interface device physically connects users with the robot, and the motion of the user is recognized and delivered to the robot by the interface device in order to provide the necessary physical force to the user. By using the force sensor that is built inside the interface device, the intention of the user’s motion is recognized. Although the sensor is capable of measuring the magnitude of force, it is unable to distinguish between the force applied by the exterior and the external force generated by the user. Moreover, in case of the electromyogram signal, it is only expressed when the user actually generates the force, whereas its exact magnitude cannot be measured. Therefore, the suggested algorithm includes a step for each of the following: filtering the signal extracted from the electromyogram sensor, determining the recognition of the user’s motion after the filtering, classifying the signal expressed by the force sensor on the orthogonal coordinate space in accordance with the user’s motion, and outputting the user’s motion intention signal by converting the classified signal into a signal, related with the force and torque. In other words, the algorithm lets the user’s motion intention signal measured by the force sensor to be output only when the user’s motion is recognized as the electromyogram signal. According to the suggested method— which relies on eliminating the signals that are output, except for the motions generated by the user—the reliability of the user motion intention signal delivered to the robot can be enhanced.