Development of a wearable multi-sensor system for lower limb joint torque evaluation during stairs climbing

Torque control is significant for the control of the exoskeleton robot. Different from the classical Newton-Euler method, this paper introduces a novel human model during low-speed stairs climbing for estimating the lower limbs joint torque in the sagittal plane. In this model, the stair ascent gait cycle is divided into four phases: single support, double support, right leg support with left redundancy and left leg support with right redundancy, which manifest to four different models. To verify the model, IMUs and smart shoes are used to measure the joint kinematic and ground contact force. A fuzzy logic rule is proposed for gait analysis in one stair ascent gait cycle. The optical measurement system and force plates are used as a reference, of which the basic principle is the Newton-Euler method. The results of kinematic and dynamic analysis based on the model are compared with the reference through a low-speed climbing stairs experiment with different loads.