Human motion recognition systems play an important role in improving the ability of robots to support human life. This paper proposes a novel motion recognition method based on haptic data (i.e., position and force information). First, haptic data are obtained using a bilateral control system. Second, the haptic data are divided into action components, using modal transformation. Finally, both the position and the force information in each action component are compared with recorded data. A dynamic programming pattern-matching algorithm is used to recognize the desired motion, and the validity of the proposed method is verified experimentally. In the experiments, the proposed method is applied to a grasping motion. The proposed method can trigger scaled bilateral control and assist the operator in real time.
ASJC Scopus subject areas
- コンピュータ サイエンスの応用