MIT researchers have created a new memory system called UQ-DAAAM (Uncertainty Quantification for Describe Anything, Anywhere, at Any Moment) that identifies entries in a robot’s memory that might be wrong and corrects descriptions of mislabeled objects, which could help household or industrial robots perform complex, multistep tasks such as clearing a dining room table or assembling machinery alongside human collaborators. The system compares multiple descriptions of the same object seen from different viewpoints to quantify the robot’s uncertainty about what it saw, then selects additional views to reduce that uncertainty and combines the captions into a more accurate description saved to memory. The researchers employed a special technique to streamline the process so a mobile robot can continually update uncertain memory entries in real time. Compared to baseline methods, UQ-DAAAM boosted the robot’s confidence and improved the accuracy of its answers to questions about objects, their locations, and when it saw them. The work builds on prior DAAAM, a robot memory system using a vision-language model (VLM) to describe objects captured with onboard cameras and store them as captions in a 3D map-based representation. Luca Carlone, an associate professor in MIT’s Department of Aeronautics and Astronautics (AeroAstro), principal investigator in the Laboratory for Information and Decision Systems (LIDS), director of the MIT SPARK Laboratory, and senior author of the paper, is joined by co-lead authors and MIT graduate students Harry Zhang and