A Vibrotactile Bracelet for Emergency Alerts for the Deaf Community in Day-To-Day Life

Abstract

Emergency alert systems that rely solely on auditory signals pose significant risks to the Deaf and Hard-of-Hearing (DHH) community, especially in public and crowded environments. In response to this accessibility gap, this project focuses on the design and implementation of a vibrotactile emergency alert bracelet that delivers non-auditory feedback using real-time environmental sound recognition. The bracelet integrates an Arduino Nano RP2040 Connect microcontroller, which features an onboard MP34DT06JTR MEMS microphone (Arduino, n.d.), and employs a mini vibration motor and OLED display to provide tactile and visual alerts. Emergency sound types, including fire alarms, sirens, and public announcements, are classified using machine learning models trained with Edge Impulse. The vibration feedback is controlled through Linear Resonant Actuators (LRAs), chosen for their efficient, low- power haptic performance in wearable devices. Feature extraction is performed using Mel-Frequency Cepstral Coefficients (MFCC), and classification models are evaluated based on accuracy, latency, and robustness to untrained samples. The system was validated through real-world testing, and results demonstrate high classification accuracy for tonal alerts and effective user recognition of vibration patterns. Limitations remain in detecting speech-based announcements. Battery drains tests and user surveys confirm the system’s reliability for daily short-term usage. This project presents a cost-effective, wearable solution that enhances situational awareness and safety for the DHH community in emergency scenarios.