Adaptive Stair-climbing Behaviour with a Hybrid Legged-Wheeled Robot

Markus Eich, Felix Grimminger, Frank Kirchner

In: L. Marques, Almeida, M.O. Tokhi, G.S. Virk (editor). Advances in Mobile Robotics. 11th International Conference on Climbing and Walking Robots and the Support Technologies for Mobile Machines. International Conference on Climbing and Walking Robots and Support Technologies for Mobile Machines (CLAWAR-2008) September 8-10 Coimbra Portugal Pages 768-775 ISBN 978-981-283-576-5 World Scientific Publishing Co. Pte. Ltd. Danvers, MA, USA 2008.


Inspired by quadruped animals we developed the hybrid legged-wheeled robot Asguard. We showed already that this robot is able to cope with a variety of stairs, very rough terrain, and is able to move very fast on flat ground. We will describe a versatile adaptive control approach for such a system which is based only on proprioceptive data. An additional inclination and roll feedback is used to make the same controller more robust in terms of stair-climbing capabilities. At the same time, high velocities can be reached on flat ground without changing the configuration of the system. By using twenty compliant legs, which are mounted around four individually rotating hip-shafts, we abstract from the biological system. For the locomotion control we use an abstract model of bio-inspired Central Pattern Generators (CPG) which can be found in biological systems from humans to insects. In contrast to existing work, Asguard uses the sensed feedback of the environment to adapt the walking pattern in real time.

German Research Center for Artificial Intelligence
Deutsches Forschungszentrum für Künstliche Intelligenz