Objective: This study presents the design of an innovative lower limb exoskeleton featuring dynamic self-balancing capabilities. It employs Zero Moment Point and Model Predictive Control online gait algorithms to plan stable walking patterns, thereby facilitating precise gait training for individuals with lower limb motor impairments and enhancing the overall training effectiveness. Methods: The positive and negative kinematic solutions of the lower limb exoskeleton were determined using the Denavit-Hartenberg method and the geometric method, respectively. The geometric relationships of the exoskeleton’s linkage components were employed to derive workspace expres sions for various gait phases. By utilizing Zero Moment Point and Model Predictive Control online gait algorithms, simulation experiments were conducted to validate the dynamic self-balancing capability of the exoskeleton while walking on flat terrain. Results: In the evaluation of the online gait generation algorithm’s validity, the generated gait trajectory aligned with the planned trajectory. When examining dynamic self-balancing walking capability, the trajectories from initial simulation experiments on flat terrain closely matched the intended trajectories. Conclusion: The online gait generation algorithm presented in this study is capable of producing a stable walking pattern for continuous bipedal gait. This newly designed lower limb exoskeleton can achieve stable dynamic self-balancing walking.
Keywords: Lower limb exoskeleton, online gait generation algorithm, dynamic self-balancing, rehabilitation training