Pressure-Dependent Facial Expression Control Using Calibrated Force-Sensitive Sensors

This study presents a compact affective sensing system that converts physical touch into intuitive multimodal feedback through dynamic facial expressions and optional audio responses. The system was developed to support non-verbal, human-centered interaction in embedded human–machine interfaces, where tactile input can be directly associated with emotional perception. The hardware platform is based on the M5Stack CoreS3, integrating a force-sensitive resistor (FSR), an ESP32-S3 microcontroller, an embedded LCD, and a built-in speaker. Pressure signals are acquired using a simple voltage divider circuit and digitized through the built-in 12-bit analog-to-digital converter (ADC) of the ESP32-S3. To improve signal stability, a simple moving average (SMA) filter is applied for noise reduction. The normalized pressure signal is classified into multiple pressure regions and mapped to emotional states. Smooth facial transitions are generated by continuously interpolating geometric facial parameters, including mouth curvature, eye shape, and eyebrow angle, without relying on pre-rendered images. Experimental evaluation demonstrated stable pressure detection, low-latency response, and intuitive emotional feedback across a wide operating range. User evaluation results further indicated that the combination of visual and auditory feedback enhanced realism, anthropomorphic perception, and interaction quality. The proposed system demonstrates the potential of tactile affective sensing for applications in assistive communication, education, and empathetic human–robot interaction systems.