MIT researchers develop uq-daam to correct robot memory
mit researchers introduced uq‑daam, a memory system that identifies and corrects mislabeled object descriptions in robots. the system uses uncertainty quantification to detect when a robot’s vision‑language model (vml) is unsure about a caption and then gathers additional viewpoints to refine the description. the method was tested on a mobile robot that navigates indoor scenes.
Key points
- uq‑daaam uses uncertainty quantification to detect and correct mislabeled robot memory, improving description accuracy by 5 %.
- the system resolves ambiguities for about 92 % of objects with only two additional captions per item.
- mit researchers plan to apply uq‑daaam to mobile manipulators and use uncertainty maps to guide re‑visits.
in experiments, uq‑daam improved the accuracy of object descriptions by 5 % over the previous daaam system and resolved ambiguities for about 92 % of objects while adding only two extra captions per item. the authors say the approach can run in a fraction of a second on a moving robot, making it practical for real‑time updates. they plan to extend the technique to other robot types and to use uncertainty maps to guide re‑visits of uncertain areas.
the work is supported by the u.s. army research laboratory and the u.s. office of naval research. luca carlone, the senior author, is an amazon scholar on sabbatical. the paper will appear at the conference on neural information processing systems.
Clearing up a robot’s fuzzy memory
MIT News on AI · 5 November 2026
Loading the full article…
This text was published by MIT News on AI and written by Adam Zewe | MIT News. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Research
All →- Anthropic uses Claude to map bacteriophage vault protein at 3.2 Å resolution · 3 src
- RAGCompass routes retrievers by reranking passages, improves answer F1 by 28% · 1 src
- Qwen improves LLM-based OCR correction with veto gate · 1 src
- Study: LLMs prefer to answer over clarifying ambiguous references · 1 src
- Researchers release SRRM benchmark to test small language models’ Long-Context memory · 2 src
Comments
via GitHub Discussions