Robotic vertical jumping agility via series-elastic power modulation

TLDR

Arboreal mammals can repeatedly jump up to 2 m, and the galago achieves the highest vertical jumping agility by using power‑modulating strategies, yet most robots with series‑elastic power modulation reach only about 55 % of the galago’s agility. The study defines and characterizes vertical jumping agility as a metric for rapid, repeated vertical jumps. Kinetic analysis reveals that vertical jumping agility is fundamentally limited by the actuator power available. Using a specialized leg mechanism that enhances power modulation, we built a robot that attains 78 % of the galago’s vertical jumping agility, demonstrates a wall‑jump exceeding a single jump, and shows that series‑elastic power modulation enables vertically agile robots.

Abstract

Several arboreal mammals have the ability to rapidly and repeatedly jump vertical distances of 2 m, starting from rest. We characterize this performance by a metric we call vertical jumping agility. Through basic kinetic relations, we show that this agility metric is fundamentally constrained by available actuator power. Although rapid high jumping is an important performance characteristic, the ability to control forces during stance also appears critical for sophisticated behaviors. The animal with the highest vertical jumping agility, the galago (Galagosenegalensis), is known to use a power-modulating strategy to obtain higher peak power than that of muscle alone. Few previous robots have used series-elastic power modulation (achieved by combining series-elastic actuation with variable mechanical advantage), and because of motor power limits, the best current robot has a vertical jumping agility of only 55% of a galago. Through use of a specialized leg mechanism designed to enhance power modulation, we constructed a jumping robot that achieved 78% of the vertical jumping agility of a galago. Agile robots can explore venues of locomotion that were not previously attainable. We demonstrate this with a wall jump, where the robot leaps from the floor to a wall and then springs off the wall to reach a net height that is greater than that accessible by a single jump. Our results show that series-elastic power modulation is an actuation strategy that enables a clade of vertically agile robots.