A Behaviour Network Concept for Controlling Walking Machines

The high complexity of the mechanical system and the difficult task of walking itself makes the task of designing the control for legged robots a diffcult one. Even if the implementation of parts of the desired functionality, like posture control or basic swing/stance movement, can be solved by the usage of classical engeneering approaches, the control of the overall system tends to be very unflexible. This paper introduces a new method to combine apects of classical robot control and behaviour based control. Inspired by the activation patterns in the brain and the spinal cord of animals we propose a behaviour network architecture using special signals like activity or target rating to influencce and coordinate the behaviours. The general concept of a single behaviour as well as their interaction within the network is described. This architecture is tested on the four-legged walking machine BISAM and experimental results are presented. keywords: Four-Legged Walking Machine, Behaviour Based Control