主催: 一般社団法人 日本機械学会
会議名: ロボティクス・メカトロニクス 講演会2017
開催日: 2017/05/10 - 2017/05/13
We aim to make a one-legged hopping robot hop continually by employing a neural network-based controller, which is trained through remote balancing control by humans. Consequently, the height and stability of the hop become important for human control of the robot's balance. Our robot uses an active-suspension mechanism. The robot performs dumping with motors at foot landing and exerts the instantaneous power of a spring to jump. However, it is difficult to achieve a stable jump with sufficient height because the spring and the motor are mutually constrained. Therefore, we introduced a mechanical or electromagnetic clutch mechanism into the robot, and succeeded in increasing power of the jump. In this study, we re-examined the spring and the motor in terms of their abilities to improve the quality of hopping. In addition, we introduced a new control that reduced the preparation time for the next jump and restrained rebounding at landing.