Reinforcement Learning for Minimizing Age of Information under Realistic Physical Dynamics | IEEE Conference Publication | IEEE Xplore