A comparison of different methods for modelling the physical human-exoskeleton interface

There are several methods to simulate the human-exoskeleton interface but there is insufficient evidence regarding the choice of the method. This work compares two rigid-body methods to simulate the interface: 1) optimisation-based contact forces; 2) reaction forces at a point on the interface. Additionally, a method to kinetically align the human-exoskeleton joint axes is presented. A single subject tested an active lower limb exoskeleton in stair ascent. The biomechanical outputs were compared to a baseline model, where the measured assistive and ground reaction forces were applied directly to the human model. Both methods showed negligible differences in knee compression force, knee flexion moment, and vastus lateralis activation. However, the ankle outputs showed some differences between the methods. Computationally expensive contact forces provided six-axis interface forces unlike reaction forces, which were limited to the number of constraints required by the exoskeleton. Future studies could compare rigid-body and viscoelastic models.