Research on Human-machine Synergy Control Method of Lower Limb Exoskeleton based on Multi-sensor Fusion Information

Human walking movement is realized by bilateral lower limb height coupling. The unilateral lower limb dysfunction imposes great limitations on the patient’s normal activities. This paper addressed the rehabilitation training needs of patients with unilateral lower extremity disabilities, which utilized the normal motor function of the healthy limb to guide the affected lower limb for coordinated movement. The lower limb exoskeleton (LLE) facilitated coordinated movement in five gait modes by utilizing the multi-sensor fusion information (MSFI) set. The paper aimed to tackle the crucial challenges related to intention recognition, trajectory following, and motion control of LLEs during rehabilitation training tasks. A novel MSFI set for LLE, based on a combination of low-cost sensors, was proposed in this paper to enhance the accuracy of attitude estimates. To further improve the accuracy and timeliness of the model, a spatiotemporally embedded convolutional long-short memory deep learning network (TSLSTM) was introduced. Additionally, a double closed-loop (DCL) controller was designed to address poor human-machine synergy issues. Experimental tests, including trajectory following and stable walking experiments, were conducted to validate and evaluate the proposed model. The results offered valuable insights for addressing intent recognition and enhancing human-machine collaboration for LLEs.