Terrain-Adaptive Gait Planning for Lower Limb Walking Assistance Exoskeleton Robots

In recent years, Lower Limb Exoskeletons (LLEs) have gained considerable interest walking assistance applications for paraplegic patients. Assisting patients to walk over different terrains such as slopes and stairs is necessary for the widespread use of LLEs. In this paper, a Kernelized Movement Primitives (KMP) based terrain adaptive gait planning approach has been proposed, which can be used to generate adaptive gait trajectories online and dynamically. The main idea is to model the human-exoskeleton system as a reduced order dynamics model, with the given step length, the foot positions can be generated, and the gait trajectory can be encoded and reproduced by the KMP with the geometric parameters of the terrains that measured by the vision sensor. After learning from demonstrated gaits that sampled from healthy subjects, adaptive gait trajectories can be reproduced on-line to adapt to slopes and stairs with different geometric sizes. The efficiency of the proposed approach has been demonstrated on an lower limb walking assistance exoskeleton robot, and experimental results indicate that the proposed approach can provide exoskeletons with the ability to generate appropriate gait trajectories for different terrains.