Multi-Fidelity Reinforcement Learning with Gaussian Processes

We study the problem of Reinforcement Learning (RL) using as few real-world samples as possible. A naive application of RL can be inefficient in large and continuous state spaces. We present two versions of Multi-Fidelity Reinforcement Learning (MFRL), model-based and model-free, that leverage Gaussian Processes (GPs) to learn the optimal policy in a real-world environment. In the MFRL framework, an agent uses multiple simulators of the real environment to perform actions. With increasing fidelity in a simulator chain, the number of samples used in successively higher simulators can be reduced. By incorporating GPs in the MFRL framework, we empirically observe up to $40\%$ reduction in the number of samples for model-based RL and $60\%$ reduction for the model-free version. We examine the performance of our algorithms through simulations and through real-world experiments for navigation with a ground robot.