Low-Force Physical Human-Robot Interaction at the Hands Influence Changes to Gait Coordination through Sensorimotor Engagement Instead of Direct Gait Propulsion

Abstract

By influencing human walking through sensorimotor engagement, low-force hand interaction has the potential to improve safety, efficacy, and convenience of gait assistance/rehabilitation and human-robot collaborative tasks. However, low-force physical human-robot interaction (pHRI) at the hands has been rarely studied in the context of influencing walking, especially for changing gait coordination, an important skill that is impaired with aging and disease. To develop low-force hand pHRI that is both effective and easy to use, we must understand how humans respond to the interaction intuitively, without explicit instructions or training. We conducted experiments using a versatile emulator robot to examine our hypothesis that low-force hand pHRI influences changes to gait coordination through human sensorimotor engagement instead of providing gait propulsion. Indeed, only participants expecting task-relevant information from pHRI changed gait coordination, and mechanical power transfer at the hands was insufficient to propel walking. Without explicit instructions, participants intuitively rejected robot motions with arm effective impedance. Our low-force hand interaction paradigm facilitates study of physical interaction separate from the primary physical task of walking, and our results suggest design principles for a new type of robotic walking aid to influence gait through human sensorimotor engagement.