Task-Driven Detection of Distribution Shifts With Statistical Guarantees for Robot Learning

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-Bayes theory to train a policy with a guaranteed bound on perfo...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE transactions on robotics Jg. 41; S. 926 - 945
Hauptverfasser: Farid, Alec, Veer, Sushant, Pachisia, Divyanshu, Majumdar, Anirudha
Format: Journal Article
Sprache:Englisch
Veröffentlicht: IEEE 2025
Schlagworte:
Collision avoidance
Costs
Deep learning in robotics and automation
Detectors
Drones
failure detection and recovery
formal methods in robotics and automation
Grasping
Hardware
Navigation
PAC-Bayes
Robot sensing systems
Robots
Training
ISSN:1552-3098, 1941-0468
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung: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-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 <inline-formula><tex-math notation="LaTeX">p</tex-math></inline-formula>-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.
ISSN:1552-3098
1941-0468
DOI:10.1109/TRO.2024.3521963