robotics
Failure Detection Watched Relationships, Not Whole Scenes
RAFAIL reached 73.4% balanced accuracy across three real-world manipulation tasks without failure training data.
Summary
RAFAIL reached 73.4% balanced accuracy across three real-world manipulation tasks without failure training data.
Whole-scene anomaly detectors can react to harmless visual changes, while runtime vision-language models add computation. RAFAIL learns representations of task-relevant relationships—such as gripper to object or object to target—from successful demonstrations annotated offline. At runtime, relationship-specific out-of-distribution detectors run without VLM inference or failure examples. Across three real-world manipulation tasks, the method reached 73.4% balanced accuracy and outperformed the strongest evaluated uncertainty and OOD baselines. The result supports narrowing failure detection to the geometry that matters for task progress.
Why it matters
RAFAIL reached 73.4% balanced accuracy across three real-world manipulation tasks without failure training data.
Limits and context
No additional limitation was separately recorded.
Key claims
RAFAIL reached 73.4% balanced accuracy across three real-world manipulation tasks without failure training data.
Evidence: source-2026-09-17-011
Sources
- arXiv preprint 2609.18324arXiv · primary research
Corrections
No corrections have been recorded for this story.