Gait rehabilitation training robot: A motion-intention recognition approach with safety and convenience

Abstract

Motion-intention recognition is a vital prerequisite in active training when employing a gait rehabilitation training robot. To accurately recognize the motion intention of the elderly and the people with inconveniences in the human–robot interaction process, a novel approach of motion intention recognition with safety, accuracy, and convenience, including directional intention recognition (DIR) and speed intention recognition (SIR) is proposed in this paper. Firstly, the structures of the gait rehabilitation training robot and its motion-intention recognition system are illustrated. To ensure that the user walks in any desired direction safely and naturally, an improved distance-type fuzzy reasoning algorithm combined with a shake-intent filter and second-order optimization algorithm is proposed. It effectively eliminates the control error caused by body shaking and usage habits in human–robot interaction. Furthermore, by extracting from the pressure sensor data, a novel algorithm, taking advantage of the Gaussian probability density function (PDF)’s characteristics, is proposed for SIR, which does not increase the system complexity. Finally, a multi-directional fuzzy reasoning experiment and a human–robot interaction experiment are implemented. The results show that the algorithm can accurately recognize the motion direction intention and motion speed intention of people with weak motion capability, which also improves the safety and convenience of the interaction approach. The proposed method can be integrated into a walker with similar structures, and the whole system can be applied in hospitals, families, and other places for assisting the elderly and the disabled.