Direct User Control of Knee Exoskeletons

Assistive lower limb exoskeletons have the potential to improve the quality of life for people with mobility disabilities as well as the endurance of first responders in emergencies. They can do this by inferring gait phase from motion sensing and applying a corresponding torque profile—typically a scaleddown version of the biological torque profile—to relate gait phase to the appropriate assistive torque at any instant in time. However, we have no guarantee that the scaled biological torque profile is the most desirable behavior an assistive device with limited torque can provide. Therefore, we propose a direct user control framework that allows individuals to modulate their torque assistance in real time through spring-loaded analogue input buttons. The customized torque profiles that result from this direct human control become our model of the human’s desired exoskeleton behavior. In an able-bodied participant study ($N=10$), we empirically measure the ratio of 1 -norm swing and stance torques in these custom profiles and compare with the biological profile. The result shows that what users want from a bilateral knee exoskeleton is quite different from what typical controllers give them ($p \leq 0.001$), and suggests broadening the search for useful assistive profiles.