Model-based Reinforcement Learning with Parametrized Physical Models and Optimism-Driven Exploration | Read Paper on Bytez