Reinforcement learning is a natural candidate for integration with behaviour-based robot control because it seems well suited to the problems faced by autonomous agents. However, previous attempts to use it in mobile robots have been simple combinations of these two methodologies rather than full integrations, and have suffered frm severe scaling problems that appears to make such combinations infeasible. Furthermore, the implicit assumptions that form the basis of reinforcement learning theory were not developed with the problems faced by autonomous agents in complex environments in mind. In this presentation I will introduce a model of reinforcement learning that is designed specifically for use in behaviour-base robotics, taking the conditions faced by situated agents into account. This model layers a distributed and asynchronous reinforcement learning algorithm over a learned topological map and a standard behavioural substrate. This is intended to make reinforcement learning feasible and compatible with behaviour-based design principles. I will also briefly outline the development of Dangerous Beans, a mobile robot that makes use of the model, and the results of an experiment in which Dangerous Beans is required to perform puck foraging in an artificial arena. The results show that the model is able to learn rapidly in a real environment. I will also cover some of the implications of these results for further research.