A Soft Tissue and Sensor Model of Exoskeletons for Amplifying Astronaut Strength

Long periods in microgravity reduce strength, and moving from the Moon to Mars will only exacerbate this challenge for human spaceflight. Countermeasures for muscle atrophy, already a key component of astronaut safety and health, do not fully prevent muscle atrophy and loss of strength (mean isokinetic strength still decreases 8–17 % over 163±38 days of spaceflight). Assistive exoskeletons have a unique potential to offer a two-part countermeasure for atrophy: in space they could resist human motion as an exercise tool, and in planetary operations they could artificially increase human strength as an assistive device. Yet, while today's assistive exoskeletons are controlled to match the periodic motion of the lower-body in walking, which has been successful in reducing metabolic cost of transport (for example, by 17.4 % with a hip exoskeleton), this pattern-based control is unlikely to be suitable for space flight and planetary operations where motion may become more irregular. Thus, we seek a task-invariant exoskeleton controller that makes no such assumptions with respect to human behavior. One such control strategy, which we term Force-Feedback Strength Amplification (FFSA), appears ideally suited for this type of unpredictable, non-periodic locomotion. By directly measuring the effort exerted by the human operator (the force-feedback), this strategy provides proportional assistance (amplifying strength) with no need for task-knowledge. However, FFSA is sensitive to the dynamics of the human-machine interface, especially the compliant tissue dynamics and unconscious human impedance that could result in operator-induced instability if not properly modeled and compensated for in control design. In this paper, we propose a robotic testbed to test an alternative approach to measuring the elements necessary for feedback control. In our approach, only the Ground Reaction Force (GRF) vector and the relative pose (i.e., position and orientation) of an actuator and six-axis force sensor are used for feedback, which measures the human-environment interaction rather than the human-exoskeleton interaction. An active LED camera tracking system provides the relative pose measurement, while the six-axis force/torque plate provides the GRF vector. Additionally, a physical model of the human leg is used to mimic the compliance of human soft tissue using 3-D printed thermoplastic polyurethane. System identification experiments with variable compliance were conducted to validate the hardware and observe the open and closed-loop transfer functions of the system. The experiments showed strength amplification of the closed-loop controlled exoskeleton at low frequencies and an unexpected dynamic behavior at high frequencies. These data will help to design exoskeleton hardware capable of high bandwidth strength amplification in space applications.