Today’s terrestrial means of transportation and exploration are mainly based on wheeled machines. Nevertheless, about half of our planet’s land mass is impassable for wheeled machines. Walking machines can contribute exploiting areas that are unexploited so far. Walking machines bring important advantages along based on their mobility as well as on their adaptivity regarding the ground’s changing properties. Scientists have been researching on walking machines for decades. Although biological walking organisms are all over the curiosity for biological organisms as model for walking machines was stirred up about twenty years ago. Studies on biological organisms like insects became more interesting for scientists doing research on the principles of walking. Whereas in the first years the neuronal connections and the underlying control structures for leg movements and leg coordination were very important, during the last years the specific characteristics of muscles, sinews, and joints became more and more important for the biological organism’s adaptive walking. In this thesis a decentralised, reflex-based control structure for walking with a six-legged robot is introduced and its adaptivity, its dependability as well as its robustness are analysed. The thesis is based on the OCRA-concept developed within the DFG priority program Organic Computing. Following the walking pattern’s introduction different reflexes supporting walking in difficult terrain are presented. Moreover, the control system’s robustness is evaluated by controlled disturbances of the walking behaviour. Besides developing a biologically inspired control structure for walking this work is focused on the control structure’s extension by additional, proprioceptive leg reflexes for overcoming obstacles as well as increasing the system’s robustness for strong disturbances. Furthermore, combined joint reflexes reacting to an external force and also generating a walking behavior are investigated. The presented concepts are evaluated by different controlled experiments and trials with the robotic platform. The results analysis, evaluation as well as their classification are based on data recorded during the trials.
|Qualification||Doctorate / Phd|
|Publication status||Published - 2010|