Clearing up a robot’s fuzzy memory
A new system identifies and corrects unreliable object descriptions stored in a robot’s memory, improving the robot’s ability to interact with its environment.
A new memory system, called UQ-DAAAM, could prevent robots from confusing objects or failing to locate them in emergency situations. MIT researchers developed this system to identify and correct mislabeled objects in a robot's memory. By comparing multiple descriptions of the same object seen from different angles, UQ-DAAAM quantifies the robot's uncertainty and refines captions to improve its memory accuracy.
The system also uses a special technique to continually update memory entries in real time, allowing the robot to perform complex tasks safely. Professor Luca Carlone, the senior author of the paper, emphasizes the importance of robots knowing what they don't know for safe deployment. Co-lead authors Harry Zhang and Nicolas Gorlo conducted the research, which will be presented at the Conference on Neural Information Processing Systems.
This new method builds on previous work, DAAAM, which uses vision-language models to describe objects in a robot's environment. UQ-DAAAM integrates uncertainty quantification into DAAAM and identifies unreliable descriptions through additional viewpoints, ultimately improving the robot's understanding of its surroundings.
Written by urgent.news from MIT News Research's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
This story
This is one outlet's version. Read the fullest account.
- Clearing up a robot’s fuzzy memory news.mit.edu
