A Logarithmic Barrier Method For Proximal Policy Optimization