Evolving robot controllers in PDL using genetic programming
Author
Shannon, P.
Nehaniv, C.L.
Attention
2299/6952
Abstract
We demonstrate how autonomous agents can be evolved whose controllers are programs in the dynamical systems-oriented, behaviour-based PDL language introduced by Luc Steels. Controllers evolved using a physics simulator solved the problem of parallel parking a car without colliding into obstacles on three sides of a parking area.