robot learning is often difficult due to the expense of gathering data. The
need for large amounts of data can, and should, be tackled with effective
algorithms and leveraging expert information on robot dynamics. Bayesian
reinforcement learning (BRL), thanks to its sample efficiency a