History of Robotics Research and Development of Japan2012Integration, Intelligence, etc.EMG-to-Motion Classification for Prosthetic Applications - A Self-Organizing Approach with Level of Proficiency -

Kahori KitaChiba University
Ryu KatoYokohama National University
Hiroshi YokoiThe University of Electro-Communications
The purpose of this research is to enhance the control of prosthetics by discriminating between different types of motions from their EMG signal. The basic idea of this study is that the current degree of proficiency, i.e., how well the prosthetic is operated by a user, can be analyzed from bio-signals closely related to human motor processes, and a control rule can be generated depending on proficiency level. Thus, our proposed system consists of two parts: estimation of the user’s proficiency level using the EMG signal, and motion classification using self-organized clustering (Fig. 1). In the experiment, users trained to discriminate motions using our proposed system, and all were able to discriminate seven forearm motions with approximately 90% accuracy. Additionally, one of the users was able to discriminate nine motions with 21% higher accuracy than before training. The results indicated the effectiveness of proposed system. 26th RSJ Best Paper Award in 2012.
Overview of the proposed system. It consists of three phases: training phase, learning phase, and pattern recognition phase.

Related Article