Human locomotion is part of every day's life. Nevertheless, it
is a complex motion that could not be mastered by humanoid robots
despite decades of research. How can we get a grasp of this motion?
In
this presentation, I will present how to use minimalistic gait models
(forward simulation models) to describe human walking and running.
These models do not only exhibit similar dynamics, but can also
reproduce the dynamic behavior of experimentally observed individuals
in terms of similar linearized Poincaré maps. These abstract mechanical
models can provide a common description for both biological and
artificial gait as well as guidelines how to build and control
technical walking devices like humanoids, but also prostheses and
orthoses.
In this presentation, first general characteristics of human running
are presented. Subsequently, the "linear" part of the dynamics
is identified and analyzed. Then, the notion of "template"
models is introduced, and their use in gait analysis and synthesis is
discussed. Finally, a short outlook of how
to use these models directly for controlling bipeds is given.
Universität Bonn, Institute for Computer Science, Autonomous Intelligent Systems