Cyclic Genetic Algorithms for the Locomotion of Hexapod Robots

Robotics control problems, such as gait coordination, require sequential solutions where a series of actions is continually repeated. Genetic Algorithms that do parameter optimization have not been widely applied to these cyclic sequential decision problems; although some form of evolutionary computation would be well suited for the adaptability required. In this paper we introduce Cyclic Genetic Algorithms, which were developed precisely for this purpose. The specific problem addressed, adaptive gait development for hexapod robots, was the impetus for this new kind of evolutionary computation, but it can be applied to other robotics domains.