3930 Behavior Acquisition of an Autonomous Robot Based on Reinforcement Learning with a Function of Adaptive Segmentation of the Action Space
Toshiyuki Yasuda, Kazuhiro Ohkura, Toshiharu Taura · The proceedings of the JSME annual meeting · 2005
We have been developing a new reinforcement learning called BRL. BRL has a unique feature that it not only learns in the learning space but also changes the segmentation of the learning space simultaneously. In this paper, a new function for segmenting its action space based on parameters of acquired rules is adopted to BRL in order to accelerate learning. We verify the performance of the proposed approach through computer simulations and physical experiments.