Muscle Synergy-Based Human-in-the-Loop Optimization for Personalized Hip Exoskeleton Control

Wei Yang, Zehao Yan, Linfan Yu, Luying Feng, Lianghong Gui, Canjun Yang

Abstract: Exoskeletons are promising devices used for walking training in patients with weak muscle strength. However, it is currently challenging to provide personalized assist profiles for better human-exoskeleton interaction. Human in-the-loop (HIL) optimization has been used by many studies to optimize assist torque by minimizing energy expenditure, but there are currently no studies on HIL optimization using muscle synergy as an evaluation index. In this paper, a muscle synergy-based HIL optimization framework is proposed for the hip exoskeleton to provide personalized torque assistance and reduce the effort of walking at the same time. Firstly, we put forward a muscle synergy similarity index to quantify the similarity of synergies during walking with and without exoskeleton assistance. In addition, by introducing sEMG signals to calculate the evaluation index of muscle synergy and iteratively optimizing the assist parameters in real time, muscle synergy-based HIL optimization of assist torque profile was proposed and tested on a portable hip exoskeleton. The optimal and sub-optimal torque profiles were explored through 35 iterations. The muscle synergy similarity index and the degree of muscle activation under 4 different assist profiles were tested by the validation experiment. The muscle synergy similarity during walking with exoskeleton assistance can be improved as much as possible by using the assist parameters generated by HIL optimizing, which reached 0.7113 in our experiment. This method can also reduce muscle activations of vastus medialis, gluteus maximus, tibialis anterior, soleus muscle, which proves that the HIL optimization based on muscle synergy proposed can reduce the effort of walking and ensure a “natural‟ walking as much as possible.