Posts Tagged ‘robocup2D’

Optimal final_third policy in Robocup 2D using HyperNEAT

December 9th, 2015 No comments

Final project of the Neuroevolution and Generative and Developmental Systems class ( Inspired on this paper a new policy was developed to instead of players keep the ball, players learn to score. Even though we run just 30 generations which is considered low, the best individual performed better than taking random decisions and better than a scripted behavior as can be seen not probably in the video but on the statistics gathered.