H. Ulbrich; "Walking Machines at TUM", 10.05.2005, 13:40, L063
Walking Machines at TUM
Heinz Ulbrich
Institute of Applied Mechanics, Technical University of Munich
The movement of walking machines is a complicated process that cannot be controlled by applying strategies known from other technical systems straightforwardly. Walking can be separated into different phases like single and double limb stance which can in turn be subdivided into smaller phases like loading response, mid stance, terminal stance etc. Similarly to the walking motion the system dynamics can be divided into different parts like upper body dynamics, leg dynamics and the dynamics of the single foot, which play an important role. The performance of walking machines is still far from that of biological systems. Biological control adapts perfectly to environmental changes and requirements while the systems as a whole are self-learning and even self-repairing. One of the reasons for the excellent performance of animals and humans in this area is their excellent sensory perception which still is not fully understood. As mentioned above, the sequence of events during walking cannot be controlled using classical control strategies. This is also due to the lack of sufficiently accurate and fast sensors, which are necessary for applying such strategies to the control of walking machines. However, biological systems obviously do not use such strategies either, rather relying on redundant sensors, sophisticated sensor fusion methods and other not yet fully understood mechanisms. Technical walking control systems are usually organized in a hierarchical manner with more intelligent, less specialized layers controlling the entire system and more specialized, less intelligent layers controlling individual components. In trying to improve autonomous task planning capabilities, highly developed vision systems play an important role.
Both biological and technological walking require a huge amount of logical decisions which depend on the state of the walking situation and are activated both in normal and disturbed walking.
The presentation will focus on two-, six- and eight-legged walking machines developed at the Institute of Applied Mechanics, Technical University of Munich. The realization of such walking machines including strategies for the mechanical and control system design will be presented, including some short video clips showing the performance of these walking machines.
References
WEIDEMANN, H.-J. (1993). Dynamik und Regelung von sechsbeinigen Robotern und natürlichen Hexapoden. Fortschrittberichte VDI Reihe 8, Nr. 362. VDI-Verlag, Düsseldorf
PFEIFFER, F. ; ROßMANN, TH. AND STEUER, J. (1999). Theory and Practice of Machine Walking. In CISM Courses No. 375 (Human and Machine Locomotion), eds. Morecki and Waldron. Springer Verlag Wien, New York
PFEIFFER, F. ; LÖFFLER, K. AND GIENGER, M. (2002). The Concept of Jogging Johnnie. Proc IEEE Int Conf on Robotics and Automation (ICRA), Washington DC, pp. 3129-3135
PFEIFFER, F. ; LÖFFLER, K. ; GIENGER, M. AND ULBRICH, H. (2004). Sensor and Control Aspects of Biped Robot "Johnnie". International Journal of Humanoid Robotics (IJHR), Vol. 1, No. 3, pp. 481-496
MAY 10, 2005 TUESDAY, 13.40, L063