A Balance Control Strategy Based on the CP and DMP for Lower Limb Exoskeleton

While healthy users can walk in an exoskeleton without external assistive devices, balancing walking can be a difficult task for people with neurological conditions such as spinal cord injuries (SCI). The key issue, especially when subjected to external perturbations, is how the exoskeleton generates the appropriate gait to maintain the stability of the system after the human-exoskeleton system has been perturbed by external forces. To generate a human-like gait in order to make wearing more comfortable. In this paper, we propose an adaptive gait planning method that can maintain system stability by varying the step length depending on the magnitude of the perturbation to which the human-exoskeleton system is subjected. We model the human-exoskeleton system as a two-dimensional linear inversion pendulum model (2D-LIPM). The human-exoskeleton system is perturbed by external forces in the sagittal plane, and the location of the foothold is calculated in real-time using capture point (CP) theory, followed by foot-end trajectory planning using dynamic motion primitives (DMP). The proposed method can dynamically generate the trajectory of each step of the swinging foot, after learning the gait trajectory from a normal healthy human, and can adaptively plan the anthropomorphic gait online based on the magnitude of the perturbation. We validated the effectiveness of the proposed method by simulating it on the Simulink Simscape platform. The experimental results show that the proposed method can improve the self-balancing capability of the exoskeleton robot.