Note: This content is accessible to all versions of every browser. However, this browser does not seem to support current Web standards, preventing the display of our site's design details.

  

Control of legged locomotion in challenging terrain

Back
Abstract:
I will present results from our research on autonomous locomotion control of legged robots with the goal to achieve autonomous fast and robust locomotion through very challenging and partially unknown terrain.
I will give an overlook of the locomotion control architecture for a small quadruped robot and briefly present all the central elements from high level path planning to low level differential control. Achieving robust performance requires many elements of modern computer science and engineering such as search algorithms, machine learning, and control algorithms. Interestingly no single element gives the crucial advantage. Only careful and seamless integration of all the elements and careful testing of the system on many different types of terrain is able to push performance and robustness to another level.
Then, in particular in this talk I will focus on our development of inverse dynamics methods for floating base systems, i.e. as needed for legged locomotion. Striving for overcoming the disadvantages of "traditional" high-gain position control approaches that lead to overly conservative or brittle systems in case of model uncertainties, I will show that a new floating base inverse dynamics algorithm allows us to come closer to the goal of achieving compliant locomotion and relax the requirements on terrain knowledge and precision of end-effector position planning and allows for end-effector predictive force control while walking.
I will discuss the possible generalization and application of these methods to other robots such as humanoids and how to reconcile these application driven results with results from research in biological locomotion control. Last but not least, I will give an outlook how we can combine insight from research with different platforms to arrive with truly versatile and agile robotic locomotion platforms that perform well outside the lab in very difficult terrains.

Type of Seminar:
Public Seminar
Speaker:
Dr. Jonas Buchli
Date/Time:
Jun 23, 2009   5.15 p.m.
Location:

ML F 34
Contact Person:

Hartmut Geyer
File Download:

Request a copy of this publication.
Biographical Sketch:
Jonas Buchli received a Diploma in Electrical Engineering from the ETHZ in 2003 and a Doctorate from the EPFL in 2007. His research interests include self-organization and emergent phenomena in complex systems, the theory of nonlinear dynamical systems and information concepts. And, especially the possible applications of these topics to the intersecting field of engineering and biology.