Communications of the ACM
Cheetah-Cub Quadruped Robot Learns to Walk, Trot Using Gait Patterns From Real Animal
The Cheetah-Cub robotic quadruped uses a new approach to enable it to move like a horse.
Credit: IEEE Spectrum
Researchers at the Swiss Federal Institute of Technology have developed Cheetah-Cub, a robotic quadruped based on a cheetah, which they say proves that gait primitives from the motion capture of an animal can be adapted to a robot.
The Cheetah-Cub uses a new approach to enable it to move like a horse. The team used motion-capture data showing joint trajectories of a horse walking, trotting, and galloping. From this data, the researchers extracted four kinematic Motion Primitives (kMPs), which are part of the robot's central pattern generator neural network, and developed short sequential gait cycles for each kMP.
Finding an unexpectedly low 3-percent difference between the kMPs, the researchers say "a possible interpretation is that the kMPs extracted from walk, trot, and gallop are in fact the same set of kMPs, that together are sufficient to describe the three different gaits." A single kMP may be at each gait's core, but generating a gallop from a walking gait is impossible without values to enter into the equation, so the team used mathematical transformations to separately adapt each of the gaits to the robot.
From IEEE Spectrum
View Full Article
Abstracts Copyright © 2013 Information Inc., Bethesda, Maryland, USA
No entries found