Abstract:

Biomimetic leg designs often appear to be arbitrarily chosen. To make a more objective choice regarding biomimetic leg configuration for small canine-inspired robots, we compare one hind leg to the same leg arranged in a different orientation, and show that the less biomimetic leg provides better performance. This differently-oriented leg design, which we call “trans-verse-mirrored” was more efficient and faster, both in simulation and experiment even though both leg configurations used the same passive and active components, rest angles, and monoarticular knee spring.  In experiments the normal configuration had a maximum speed of 0.33 m•s−1 and a specific resistance of 5.1. Conversely, the less biomimetic transverse-mirrored configura-tion had a maximum speed of 0.4 m•s−1 and specific resistance of 3.9. Therefore the transverse-mirrored leg’s best performance yields a 21% increase in speed and 24% decrease in specific resistance when compared to the best performance achieved in the normal biomimetic leg. The major underlying reason is that the knee spring engages more readily in the transverse-mirrored configuration, resulting in this faster and more efficient locomotion. The conclusion is that simply copying from nature does not lead to optimal performance and that insight into the role played by passive design components on natural locomotory dynamics is important.

Key words: biomimetic, legged robot, quadruped, leg design, walking, legged locomotion