Summary
Robots use special touch sensors to understand objects and interact with their environment, but different types of these sensors can confuse the robots because they send different signals. The authors created a new system called BIFTA that helps robots rapidly adjust to new and unknown touch sensors using only a small amount of new data. BIFTA works by keeping important learned features, fixing how the robot views the sensor data, and focusing on trustworthy information to improve understanding. Tests showed that BIFTA improves recognition accuracy a lot when adapting to unfamiliar sensors, making touch-sensing robots more flexible across different hardware. This method could help robots better use a variety of sensors without needing extensive retraining.
tactile sensingrobotic manipulationfew-shot learningsensor adaptationpretrained encoderspectral graphsuncertainty gatingcross-query evidenceSITR datasetfeature neighborhoods
Abstract
Advances in tactile sensing have made contact-rich perception possible, accelerating progress in robotic manipulation, material understanding, and embodied interaction. However, because optical design, elastomer mechanics, and imaging geometry differ substantially across tactile sensors, models trained on known sensor types can suffer an abrupt performance collapse on unknown sensors. To address this problem, we propose the Brain-Inspired Few-Shot Tactile Adaptation (BIFTA) framework; it draws on the brain's rapid sensory adaptation mechanism to adapt a frozen encoder to an unknown tactile sensor from a small labeled support set. BIFTA preserves pretrained representations through dual-view statistical memory, constructs support-conditioned spectral graphs to repair sensor-dependent feature neighborhoods, and applies uncertainty-gated recurrent propagation to strengthen reliable cross-query evidence. Extensive benchmarks across three tactile datasets show that BIFTA substantially improves adaptation to unknown sensors: with only 10\% labeled target data on SITR, it raises mean Sparsh accuracy from 6.86\% for the frozen source classifier to 87.09\%, exceeding the strongest implemented prior comparison by 47.22 percentage points, and these gains generalize across datasets, pretrained backbones, and tactile tasks. These results validate BIFTA for data-efficient adaptation to unknown tactile sensors and offer a promising route toward tactile models that transfer across heterogeneous hardware.