Abstract
Our goal is to perform out-of-distribution (OOD) detection, i.e., to detect when a robot is operating in environments that are drawn from a different distribution than the environments used to train the robot. We leverage Probably Approximately Correct (PAC)-Bayes theory in order to train a policy with a guaranteed bound on performance on the training distribution. Our key idea for OOD detection then 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 resulting approach (i) provides guaranteed confidence bounds on OOD detection, and (ii) is task-driven and sensitive only to changes that impact the robot’s performance. We demonstrate our approach on a simulated example of grasping objects with unfamiliar poses or shapes. We also present both simulation and hardware experiments for a drone performing vision-based obstacle avoidance in unfamiliar environments (including wind disturbances and different obstacle densities). Our examples demonstrate that we can perform task-driven OOD detection within just a handful of trials. Comparisons with baselines also demonstrate the advantages of our approach in terms of providing statistical guarantees and being insensitive to task-irrelevant distribution shifts.
Original language | English (US) |
---|---|
Pages (from-to) | 970-980 |
Number of pages | 11 |
Journal | Proceedings of Machine Learning Research |
Volume | 164 |
State | Published - 2021 |
Event | 5th Conference on Robot Learning, CoRL 2021 - London, United Kingdom Duration: Nov 8 2021 → Nov 11 2021 |
All Science Journal Classification (ASJC) codes
- Artificial Intelligence
- Software
- Control and Systems Engineering
- Statistics and Probability
Keywords
- Out-of-distribution detection
- PAC-Bayes
- generalization