$\rm PI^{{\text{2}}}$-Based Adaptive Impedance Control for Gait Adaption of Lower Limb Exoskeleton

Lower limb exoskeletons are becoming increasingly popular for aiding individuals with disabilities or limited locomotion abilities. Existing lower limb exoskeletons mainly focus on walking gait to perform medical rehabilitation training; however, the behavior patterns of patients are diverse, and a single walking gait cannot satisfy the demands of daily life. In order to promote the assisting performance of lower limb exoskeletons subject to various types of gaits, this article proposes a policy improvement with path integrals ( $\rm {PI^{2}}$ )-based adaptive impedance control strategy. Our proposed method employs a hierarchical structure. First, the zero moment point concept is employed to generate the desired angle trajectories of the hip joint and knee joint under three gaits: walking, squatting, and climbing. On this basis, $\rm {PI^{2}}$ optimization is incorporated with the impedance control to track the generated angle trajectory, wherein the impedance parameters are optimized using the deviation between the actual and desired trajectory of the lower limb exoskeleton. The human–robot interaction forces are guaranteed to remain within a small threshold even under various types of gaits. Comparative simulations and exoskeleton wearing experiments are conducted to validate the effectiveness and superiority of the proposed method.