Keeping Learning-Based Control Safe by Regulating Distributional Shift
To regulate the distribution shift experience by learning-based controllers, we seek a mechanism for constraining the agent to regions of high data density throughout its trajectory (left). Here, we present…