Genetic Network Programming (GNP) has been proposed as one of the evolutionary algorithms, which is represented by graph structures. It was extended to GNP with Reinforcement Learning (GNP-RL) which combines online learning and evolution. GNP-RL succeeded in implementing the wall following behaviors of a Khepera robot. The objective of this paper is to improve the robustness of GNP-RL by introducing fuzzy GNP with noises. Fuzzy GNP overcomes the sharp boundary problem using the probabilistic transition on fuzzy judgment nodes, which improves the exploration ability. Furthermore, the robustness of fuzzy GNP can be improved by adding Gaussian noises to the sensors during the training phase. In order to evaluate the robustness of fuzzy GNP with noises, the wall following of a Khepera robot is simulated. Simulation results show that fuzzy GNP with noises is superior to GNP-RL.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.