Back to Portfolio

Embedded Systems Design Lead - Notiphones

Jan 2025 - May 2025
Embedded CESP32-S3PCB DesignBLEPWMUSB-CHardware DesignOscilloscopeSystem Integration

Overview

Led the embedded systems design for a wearable haptic headphone module that provides vibration feedback when a wake-word is detected. The system uses an ESP32-S3 microcontroller to deliver sub-1 second latency vibration feedback, creating a seamless user experience for hearing-impaired users or those who prefer haptic notifications.

The project involved designing and fabricating a custom PCB with PWM-driven vibration motor control, RGB LED indicators for visual feedback, USB-C power management circuit, and BLE (Bluetooth Low Energy) communication for receiving wake-word detection signals from the ML model running on a companion device.

I developed low-level drivers for PWM motor control to achieve precise vibration patterns, RGB LED control for status indication, USB-C power delivery, and BLE communication stack. The system required careful bring-up and debugging of cross-team integration between the ML wake-word detection model and the hardware interfaces, ensuring reliable real-time response.

As the design lead, I also led schematic reviews, coordinated with the mechanical team for enclosure design, and managed system integration testing.

Project Context

This project was completed as part of Team 15's Notiphones project at Purdue University. The goal was to create a wearable device that provides haptic feedback when wake-words are detected, improving accessibility for hearing-impaired users.

Key Features

  • Sub-1 second latency vibration feedback
  • ESP32-S3 microcontroller for processing and communication
  • Custom PCB design with PWM motor control, RGB LEDs, and USB-C power
  • BLE communication for receiving wake-word detection signals
  • Integration with ML wake-word detection model
  • Real-time response system

Challenges

  • Achieving sub-1 second latency from wake-word detection to vibration
  • Designing efficient PWM control for vibration motor
  • Integrating ML model outputs with embedded hardware
  • Managing power consumption for wearable device
  • Cross-team coordination between ML and hardware teams
  • PCB design and fabrication for compact wearable form factor

Results

  • Successfully achieved sub-1 second latency requirement
  • Completed PCB design and fabrication
  • Integrated ML wake-word detection with hardware control
  • Led successful system bring-up and testing