3 ms·Safety-Guided Deep Reinforcement Learning via Online Gaussian Process Estimation3 points by headalgorithm 8y ago