Walking robots provide a large degree of mobility and flexibility, but demand a sophisticated motion control to guarantee fast and stable locomotion in unstructured terrain. A small introduction to the chosen behavior-based control approach, which fulfills these requirements, and its BAGEL implementation is given.
Although the proposed locomotion control is versatile and adaptive to environmental changes, a suitable parametrization is needed for optimal behavior performance in all kind of terrains.
In the presented experience-based adaptation of the locomotion control, a knowledge base is created which stores a behavior's performance in each evaluated environment and uses these experiences to adapt the robot's locomotion behavior depending on the desired action and the estimated context. Besides details of the approach, the talk includes an experimental evaluation and an outlook.