Abstract: Disclosed are embodiments for facilitating restarting operations for recovery processes in autonomous systems. In some aspects, an embodiment includes determining that an autonomous vehicle (AV) is experiencing a failure condition; identifying a plurality of restart operations to apply to one or more components of the AV contributing to the failure condition; prior to applying a restart operation of the plurality of restart operations, determining that safety conditions corresponding to the restart operation are satisfied; and applying the plurality of restart operations to the AV in accordance with an increasing order of disruptiveness of each of the restart operations to operations of the AV.

Abstract: Automating reinforcement learning for autonomous vehicles may include assigning a probability with a scenario and varying that probability based at least in part on changes in performance by the autonomous vehicle associated with that scenario. The amount of time and computational bandwidth required to train a machine-learned component of an autonomous vehicle and the accuracy of the machine-learned component may be improved by determining a reward for performance of the autonomous vehicle in a scenario based at least in part on an severity metric. The impact severity metric may be determined based at least in part on a velocity, angle, and/or interaction area associated with the impact.

Abstract: A reward determined as part of a machine learning technique, such as reinforcement learning, may be used to control an adversarial agent in a simulation such that a component for controlling motion of the adversarial agent is trained to reduce the reward. Training the adversarial agent component may be subject to one or more constraints and/or may be balanced against one or more additional goals. Additionally or alternatively, the reward may be used to alter scenario data so that the scenario data reduces the reward, allowing the discovery of difficult scenarios and/or prospective events.