Evolving Plans for the KRKa2 Chess Ending

R. Alquézar, A. Gonzalez Camargo, and A. González Romero (Spain)


Reactive planning, learning policies from examples, evolutionary computation, computer chess and artificial intelligence


The chess ending KRKa2 has been used to test how good evolutionary techniques perform in planning. Using a genetic algorithm plans were constructed and evolved; these plans intend to tell us what we should play in any KRKa2 position. The idea is to investigate whether good automatically learnt policies for a planning problem can be generated using training examples along with evolutionary algorithms. The training examples, used as input to the learning algorithm, describe specific descriptions of a number of solved instances in one domain; then to improve the learnt policies obtained from the training examples, the policies should evolve. We believe that the domain of games is well-suited for testing these ideas in planning.

Important Links:

Go Back