Design and Embedded Validation of Compact ML Models for Affective Touch Classification in a Soft Interactive Companion

Name: Design and Embedded Validation of Compact ML Models for Affective Touch Classification in a Soft Interactive Companion
Uploaded: 2026-09-16T06:01:27.027499+00:00
Duration: 405 s
Description: Researchers developed compact machine learning models for soft, interactive companions to recognize affective touch. Using a 1D CNN with 13,200 parameters, the system runs locally on an ESP32 microcontroller. A hybrid strategy combines simple rules for energetic movements with neural networks for subtle gestures, improving accuracy and power efficiency.

Researchers developed compact machine learning models for soft, interactive companions to recognize affective touch. Using a 1D CNN with 13,200 parameters, the system runs locally on an ESP32 microcontroller. A hybrid strategy combines simple rules for energetic movements with neural networks for subtle gestures, improving accuracy and power efficiency.

Open the full episode