Continuous-domain reinforcement learning using a learned qualitative state representation (2008)