Terrain-Adaptive Exoskeleton Control With Predictive Gait Mode Recognition: A Pilot Study During Level Walking and Stair Ascent

Abstract

Different gait modes and transitions correspond to different lower-limb kinetic and kinematic characteristics. To provide suitable assistance during multimodal locomotion on various terrains, finite state machine-based exoskeleton controls are widely adopted, but smooth and safe transitions between different gait modes are still challenging due to the gait mode recognition delay. In view of this, a novel terrain-adaptive, phase-based exoskeleton control is proposed in this study, which features predictive gait mode recognition and accurate gait phase estimation during gait mode transitions. Experiments in real-world terrains indicated that gait mode transitions can be reliably recognized at least 0.232 ± 0.040 gait cycle prior to the beginning of the transitional gait cycle with high accuracy (above 98.5%), enabling the exoskeleton control to predictively modulate the exoskeleton assistance and ensure the user’s safety during gait mode transitions. In addition, a pilot study during level ground walking and stair ascent was also performed in a biomechanical testing environment, and peak hip extension and flexion torques were utilized as performance criteria. Experimental results showed that the exoskeleton assistance significantly reduced the requirements for peak hip extension and flexion torques during both steady-state walking and gait mode transitions, making multimodal locomotion on various terrains less challenging for individuals with physical limitations.