An exoskeleton has to be lightweight, compliant, yet powerful to fulfill the demanding task of walking. This imposes a great challenge for the actuator design. Electric motors, by far the most common actuator in robotic, orthotic, and prosthetic devices, cannot provide sufficiently high peak and average power and force/torque output, and they normally require high-ratio, heavy reducer to produce the speeds and high torques needed for human locomotion. Studies on the human muscle-tendon system have shown that muscles (including tendons and ligaments) function as a spring, and by storing energy and releasing it at a proper moment, locomotion becomes more energy efficient. Inspired by the muscle behavior, we propose a novel actuation strategy for exoskeleton design. In this paper, the collected gait data are analyzed to identify the spring property of the human muscle-tendon system. Theoretical optimization results show that adding parallel springs can reduce the peak torque by 66%, 53%, and 48% for hip flexion/extension (F/E), hip abduction/adduction (A/A), and ankle dorsi/plantar flexion (D/PF), respectively, and the rms power by 50%, 45%, and 61%, respectively. Adding a series spring (forming a Series Elastic Actuator, SEA) reduces the peak power by 79% for ankle D/PF, and by 60% for hip A/A. A SEA does not reduce the peak power demand at other joints. The optimization approach can be used for designing other wearable robots as well.
© 2011 IEEE