High-Precision Motion Control of a Lower-Limb Wearable Robot for People with Walking Disability by Hybrid Filtered Disturbance Observer

Abstract

Control of lower-limb wearable robots has been focused on high-precision. As the motion references are optimally designed to the users, the exact realization of the intended motion is directly related to effectiveness of the assistance and safety. The exoskeletal gait assistance, however, has difficulties in achieving it as the system is involved with large uncertainties from humans. Since the disturbance observer (DOB) has shown exceptional performance in robustness, in this paper, an algorithm named gait assistive control toward practical preciseness (GACPP) is proposed. The GACPP attacks severe limitations in applying the DOB to the gait assistance; the dynamics of the exoskeletal joints varies significantly during walking as the load side alters from a free leg to the body. Moreover, the discretization process of the plant model generates certain zeros with high-frequency components, which obligates the reduction of the control bandwidth. In this paper, a system identification loop is proposed to model the joint with hybridity. Based on the identified models, a hybrid control framework including the feedforward filtering is designed. To get rid of the effects from the sampling zeros, a novel filtering process is also proposed. Finally, the GACPP is verified by experiments with the powered exoskeleton, WalkON Suit.