We utilize deep reinforcement learning to obtain favorable trajectories for visual-inertial system calibration. - View it on GitHub