Compact ML Models Classify Affective Touch in Soft Interactive Companion
Researchers describe an end-to-end pipeline for recognizing touch gestures on a plush companion robot, whose soft body and spread-out tactile sensors complicate reliable detection. The work covers the design of small machine-learning models along with embedded validation to confirm they run effectively on the device itself. The abstract is truncated, so full details of the methods and results are not yet available.