Biological systems solve problems which today's technical systems can't. The biomimetic approach to robotics is the attempt to apply solutions created by evolution to technical systems. This approach is not restricted to mechanics engineering but includes and puts emphasis on the behaviour of autonomous systems. We are convinced that the study of animal behaviour leads to models and hypotheses that help to create new technical solutions in analogy to their sucessful counterparts in nature. The longterm goal is to achieve robots that autonomously interact for an unrestricted time with a real world environment and that enhance their level of performance with experience gained from the ongoing interaction with the world. In our most challenging project, the "Scorpion Project" we are working towards this goal by integrating a low level behavioral repertoire with a high level control scheme that learns to identify temporal sequences of low level behaviour of the system. We will demonstrate how the basic ambulation patterns are designed, organised and implemented. In preliminary experiments on one of our platforms we will demonstrate the robustness of this approach. Video sequences and data derived from the machine during basic ambulation are presented and discussed in light of the next step of this project, which is to integrate these mechanisms with the higher level functional (task oriented level of control of the system.