Adaptive Control Strategy for Gait Rehabilitation Robot to Assist-When-Needed

Conventional assist-as-needed (AAN) control strategies utilize only the deviation of user's limb to a predefined trajectory, but individual condition is not considered. In this paper, the control strategy of assist-when-needed (AWN) is proposed for gait rehabilitation robotics. The robot only provides assistance in any one of the seven gait phases involving the lower-limb impairments but it works in zero force mode in the other phases during walking. A synchronous reference trajectory is generated based on the gait phases. The gait phases are detected with IMU sensors and a hidden Markov model (HMM). The level of assistance is decided independently with the deviations in the seven gait phases. Then the adaptive robotic assistance is given to assist the human walking only when needed. This control strategy has been implemented in an exoskeleton robot and tested on one healthy subject. Experiment results showed that with this control strategy, the robot significantly increased knee joint angle ranges with the existence of external resistance. The experiment results revealed that the proposed control strategy is feasible and effective in delivering flexible assistance to the subjects during overground walking.