Abstract

Lower-limb exoskeleton (LLE) is a wearable device which can provide assistance to users’ lower limb during various movements. Many LLEs have been developed to aid individuals in diverse tasks, such as rehabilitation training and movement enhancement. The control method of LLE plays a crucial role in the whole exoskeleton system. Although numerous control methods have been developed, there still exist limitations of time consumption and less adaptation. Thus, there arises a pressing need for novel intelligent control methodologies geared towards facilitating walking assistance in daily activities.

The thesis is aimed to develop an adaptive gait control methodology designed to generate adaptive gaits. Considering the human-robot interaction principle, the focus of this research is on flexible gait generation through motion primitives theory. The thesis first introduces a low-level control method based on interaction forces. An innovative gait control method based on dynamic movement primitive (DMP) is presented for adjusting gaits in real-time. An adaptive gait generation method with multimodal sensors is finally proposed in the high-level control.

The thesis contributes to the state-of-the-art of the gait control design of LLE. The integration of the AFO model with the DMP method enables real-time adjustments of users’ gaits, thereby enhancing adaptability for users. The development of an adaptive gait generation method, facilitated by multimodal sensors placed on the human body, offers more convenience by providing personalized gaits. Consequently, the proposed control strategy improves the usability of LLEs in individuals’ daily lives.