Inherent in robot control and behavior is cyclicity. Sequences of actions taken by the robot tend to repeat. This is no more apparent than in the locomotion of hexapod robots. The movement of the separate legs must be coordinated in such a way that smooth forward motion will result. The individual legs each have their own cyclic nature and the combination of leg movements forms a cycle of over all movement. In primitive robots, with simple controllers that require a string of activations for locomotion, an appropriate cycle of these activations needs to be provided. The problem is further complicated by the robot’s uniqueness and accompanying variance of leg capabilities. In previous work we introduced Cyclic Genetic Algorithms which have successfully been used to generate gaits for actual hexapod robots. In this paper, we extend on that work by showing that the CGA can adapt to disabilities in the robot and adjust the gait accordingly.
[1]
Graham F. Spencer,et al.
Automatic Generation of Programs for Crawling and Walking
,
1993,
International Conference on Genetic Algorithms.
[2]
John H. Holland,et al.
Adaptation in Natural and Artificial Systems: An Introductory Analysis with Applications to Biology, Control, and Artificial Intelligence
,
1992
.
[3]
Randall D. Beer,et al.
Evolving Dynamical Neural Networks for Adaptive Behavior
,
1992,
Adapt. Behav..
[4]
Gary B. Parker,et al.
Evolving Hexapod Gaits Using a Cyclic Genetic Algorithm
,
2000
.
[5]
Randall D. Beer,et al.
Application of evolved locomotion controllers to a hexapod robot
,
1996,
Robotics Auton. Syst..