Application of a gesture classification system to the control of a rehabilitation robotic manipulator

Abstract

This paper describes the development of a low-cost gesture measurement and recognition system employing electrolytic tilt sensors. Two methods of gesture classification by software are compared: a dynamic programming algorithm and an artificial neural network. The artificial neural network is shown to have greater classification performance when classifying degraded gestures. The gesture recognition system is employed as part of a multimodal communication platform for the control of a rehabilitation robotic system.