Evaluation on the robustness of Genetic Network Programming with reinforcement learning

摘要

Genetic Network Programming (GNP) has been proposed as one of the evolutionary algorithms and extended with reinforcement learning (GNP-RL). The combination of evolution and learning can efficiently evolve programs and the fitness improvement has been confirmed in the simulations of tileworld problems, elevator group supervisory control systems, stock trading models and wall following behavior of Khepera robot. However, its robustness in testing environments has not been analyzed in detail yet. In this paper, the learning mechanism in the testing environment is introduced and it is confirmed that GNP-RL can show the robustness using a robot simulator WEBOTS, especially when unexperienced sensor troubles suddenly occur. The simulation results show that GNP-RL works well in the testing even if wrong sensor information is given because GNP-RL has a function to change programs using alternative actions automatically. In addition, the analysis on the effects of the parameters of GNP-RL is carried out in both training and testing simulations.