發送短信: Robust reinforcement learning with Bayesian optimisation and quadrature