Event
FERRET Learns Robot Preferences Using Environment Edits
1 reports · 1 sources
Isaac Ngui, Courtney McBeth, and Nancy M. Amato propose the FERRET method, which learns robot preferences from two types of feedback: robot corrections and environment edits. Its key setup is: when a person edits the scene, the robot's motion is kept unchanged, so the only explanation for the edit is the object that was moved. The authors state that experiments show environment edits can disambiguate features and learn preferences that cannot be learned from robot corrections alone.
Report timeline
- FERRET: Using Environment Edits as a Corrective Signal for Reward Learning
The authors propose environment edits as a complementary form of feedback for robot corrections: with the robot's motion held constant, a human edits the scene so that the only explanation for the edit is the moved object. They introduce FERRET (Feature Learning from Environmental and Robot-Relational Edit Traces), which learns preferences from both robot corrections and environment corrections, and they show that environment edits can disambiguate features and learn preferences that robot corrections alone cannot.
arXiv cs.RO ·