Problem Statement

Many existing AI assistant systems are expensive, not portable, dependent on internet services, difficult to customize, and not specifically designed for accessibility. Visually impaired users and elderly people often struggle with managing reminders, tracking environmental conditions, accessing information quickly, and interacting with modern digital systems. Traditional smart assistants are often closed-source and difficult to fabricate locally. AURA addresses these challenges by providing voice interaction, spoken reminders, environmental sensing, smart notifications, a portable design, open-source development, and local fabrication possibilities.

Project Description

AURA is a portable AI-powered smart companion robot centered around the ESP32-S3 microcontroller. Users speak to the device via an INMP441 I2S microphone; the system processes commands locally or relays them to an AI API over WiFi, then responds through a MAX98357A amplifier and speaker. A TFT/OLED display shows the clock, sensor readings, battery level, and a robot face animation. A PIR motion sensor detects nearby users and triggers greetings and reminders. The entire system runs on a rechargeable Li-ion battery with a TP4056 charging module, housed inside a 3D-printed robot-inspired enclosure fabricated from PLA/PETG filament.

Target Users

Who Will Use This Device

• Visually Impaired Individuals: Voice notifications and spoken feedback make information accessible without relying on visual displays
• Elderly People: Voice reminders for medications, appointments, and daily routines in simple, accessible interaction
• Students: Affordable desktop AI assistant for study reminders, questions, and environmental awareness
• General Smart-Device Users: Portable, customizable alternative to commercial AI assistants

System Architecture

Hardware & Software Stack

The system consists of the following components:

• Microcontroller: ESP32-S3 Super Mini for WiFi connectivity, Bluetooth, I2S audio support, and AI capability
• Voice Input: INMP441 I2S microphone for digital, noise-resistant audio capture
• Audio Output: MAX98357A I2S amplifier driving a 4Ω/8Ω speaker for voice responses
• Display: TFT/OLED screen showing clock, sensor data, AI status, and robot face animations
• Sensors: DHT22/BME280 for temperature & humidity, MQ-2/MQ-135 for gas detection, PIR for motion sensing
• Power: Li-ion battery with TP4056 charging module and battery monitoring firmware
• Firmware: C++ using Arduino framework with AI API integration, NTP sync, voice synthesis, and sensor monitoring

Electronics

PCB Design & Fabrication

Custom PCB designed in KiCad, milled on the Fab Lab's CNC machine using an FR1 copper board. The board integrates ESP32-S3 headers, sensor connectors, I2S audio interfaces, power routing with a TP4056 charging circuit, user buttons, and ground planes for noise reduction and signal integrity. Each subsystem — audio, sensors, display, power — is routed with careful attention to signal separation.

Fabrication

Housing & Assembly

The enclosure is 3D-printed in PLA or PETG filament, designed to resemble a compact desktop robot. Printed components include the main shell, internal component supports, display frame, speaker housing, and battery holder. The design provides accessible ports, speaker ventilation grilles, and sensor openings. A modular internal layout allows components to be mounted securely with organized wiring and dedicated compartments, giving the finished product a professional appearance.

Programming

Firmware & Features

Firmware written in C++ for Arduino/ESP32-S3. The device operates in two modes: offline mode provides clock display, basic reminders, environmental monitoring, motion detection, and predefined voice commands without any internet connection; online mode connects to OpenAI or Gemini APIs via WiFi to answer open-ended questions and generate smart notifications. Features include automatic NTP time synchronization, I2S voice synthesis for audio responses, motion detection with greeting triggers, environmental monitoring with threshold alerts, battery level monitoring with sleep modes for power efficiency, and a robot face animation on the display.

Community Impact

Accessibility & Education

AURA addresses accessibility gaps in commercial AI assistants by providing an affordable, locally fabricable, voice-first device for visually impaired and elderly users. It serves as an educational example for learners in Fab Labs and the maker community, demonstrating the integration of embedded AI, digital fabrication, PCB design, and accessibility-centered product development. The project is especially aimed at inspiring young women in STEM. All design files, KiCad schematics, STL files, Arduino firmware, and documentation will be released as open source.

Future Work

Next Steps

Plans include developing a companion mobile application for easier configuration and remote control, integrating with smart home systems, expanding voice message options and improving audio quality, adding multilingual support for broader accessibility, upgrading to a better offline AI model as embedded hardware improves, adding a camera for face detection and recognition, and developing a DIY kit for STEM education programs targeting young women and Fab Lab learners.