A Simplified Approach to Automatic Stance Control of Unilateral Knee Exoskeletons.

For lower limb exoskeletons to be practical in real-world settings, their control systems must ensure both safety and effectiveness. A key challenge is achieving this with a simplified set of sensors, reducing system complexity without compromising performance. This paper presents a stance control algorithm that relies solely on motor current data and user kinematic information from a thigh-worn inertial measurement unit and knee encoder. The algorithm was implemented on a unilateral knee exoskeleton using a Finite State Machine to apply an automatic joint lock during stance phase of gait. The algorithm was evaluated in real-time with six healthy subjects walking on an instrumented treadmill at speeds of 0.60 m/s, 0.80 m/s and 1.00 m/s. Across subjects, no false positive or negative state transitions were observed. Locking was applied prior to heel strike, with a mean phase error of 4.81% ± 2.70%. Unlocking occurred within the desired terminal stance window, prior to toe off with a mean phase error of 5.28% ± 2.52%. These preliminary results suggest that the algorithm is effective in healthy users with varying anthropometrics and biomechanics, motivating further testing with a larger subject pool and individuals with mobility impairments.Clinical Relevance- The presented mid-level controller establishes a simple and effective methodology for automatic stance control of unilateral knee exoskeletons.