Using evolved paths for control in robot-soccer
P.J. Thomas, Russel James Stonier · 2004
In this paper we use an evolutionary algorithm to evolve paths using robot velocity profiles from an initial configuration to a configuration behind the ball. Physical conditions for wheel lift are incorporated as well as forward and reverse motion of the robot. A fuzzy clustering technique is used to build the fuzzy control knowledge base from optimised paths found for a large grid of initial configurations.