Max Planck Institute for Dynamics and Self-Organization -- Department for Nonlinear Dynamics and Network Dynamics Group
Personal tools
Log in

BCCN AG-Seminar

Wednesday, 28.09.2005 16 c.t.

Fast Biped Walking via A Reflexive Neuromuscular-like Controller

by Tao Geng
from University of Stirling, UK

Location

Seminarraum Haus 2, 4. Stock (Bunsenstr.)

Abstract

Although many biped robots have been developed using various technologies, they are still well outperformed by their natural counterpart, humans, in some important aspects such as speed and robustness. While biped robots usually depend on position control or trajectory tracking control at least in some stages of their walking gaits, humans and animals use a totally different "technology", neuromuscular control, by which stable walking gaits emerge from the global entrainment between the neuro-musculo-skeletal system and the environment. Moreover, in human and animal locomotion, the muscle plays more roles than an activator or a motor does in a walking robot. Here, we present our design and experiments of a planar biped robot and its reflexive neuromuscular control network composed of biologically plausible model neurons and a simple muscle model that is simulated with a control algorithm implemented on DC geared motors. In the experiments, our biped robot attained a relative speed that is faster than any other biped robot, and is comparable even with that of humans.

back to overview