Task-Driven Detection of Distribution Shifts with Statistical Guarantees for Robot Learning
This addresses the challenge of ensuring robot safety and reliability in unpredictable environments, though it is incremental by building on existing statistical methods.
The paper tackles the problem of detecting out-of-distribution (OOD) environments in robot learning by using PAC-Bayes theory to train policies with performance guarantees, and it shows that the approach can perform task-driven OOD detection within a handful of trials in grasping and drone obstacle avoidance tasks.
Our goal is to perform out-of-distribution (OOD) detection, i.e., to detect when a robot is operating in environments drawn from a different distribution than the ones used to train the robot. We leverage Probably Approximately Correct (PAC)-Bayes theory to train a policy with a guaranteed bound on performance on the training distribution. Our idea for OOD detection relies on the following intuition: violation of the performance bound on test environments provides evidence that the robot is operating OOD. We formalize this via statistical techniques based on p-values and concentration inequalities. The approach provides guaranteed confidence bounds on OOD detection including bounds on both the false positive and false negative rates of the detector and is task-driven and only sensitive to changes that impact the robot's performance. We demonstrate our approach in simulation and hardware for a grasping task using objects with unfamiliar shapes or poses and a drone performing vision-based obstacle avoidance in environments with wind disturbances and varied obstacle densities. Our examples demonstrate that we can perform task-driven OOD detection within just a handful of trials.