Human motion reproduction by torque-based humanoid tracking control for active assistive device evaluation

In a super-aged society such as Japan, wearable assistive devices that aim at reducing caregiver burden as well as improving the autonomy of the elderly are attracting strong interests. Humanoid robots can be used to evaluate the supportive effects of assistive devices by measuring joint torques, as that cannot be directly obtained from human subjects. While our previous work proposed a humanoid-based method for estimating static supportive torques of powerful and active supportive devices, this paper proposes a novel framework for evaluating their supportive effects during dynamic motion. Assuming that humans move with minimum exertion when taking full advantage of a device's assistive power, we propose using a controller on a humanoid to track retargeted human motions during power assistance. The effectiveness of the proposed method has been validated by experimentally assessing an assistive device (Muscle Suit) actuated by pneumatic artificial muscles.