Adaptive state construction for reinforcement learning and its application to robot navigation problems | IEEE Conference Publication | IEEE Xplore