Atmosis edge-AI environmental monitor wins Arduino UNO Q Dream Lab Challenge

Atmosis, a winning entry in the Arduino UNO Q Dream Lab Challenge co-hosted with DigiKey, reimagines the DIY environmental monitor by replacing raw sensor output with interpreted, user-friendly alerts. The device is built around the dual-brain Arduino UNO Q board and collects data from an Adafruit DPS310 pressure/altitude sensor, a Waveshare X6 multi-gas and climate sensor, and a Waveshare dust sensor. Rather than displaying PM2.5 counts or humidity percentages, Atmosis runs edge AI models via Edge Impulse directly on the board to classify whether conditions warrant user concern.

For users seeking deeper analysis, the system can optionally route data through Google's Gemini LLM to generate contextual explanations and answer follow-up questions. The project delivers output through a local web interface and a Telegram bot, packaged in a custom 3D-printed enclosure with an LED strip for at-a-glance status indication. The design specifically targets hyperlocal environments where official weather stations may not capture microclimate variations, such as construction sites or indoor workspaces.

Hardware architecture and AI pipeline

• MCU platform: Built on the Arduino UNO Q, a dual-brain board whose architecture was a core constraint and feature of the Dream Lab Challenge.
• Sensor suite: Combines three distinct I2C/SPI-connected modules: Adafruit DPS310 for barometric pressure and altitude, Waveshare X6 for gas detection plus temperature and humidity, and a Waveshare dust sensor for particulate measurement.
• Edge inference: Deploys trained models through Edge Impulse, enabling local classification of sensor fusion data without cloud dependency for basic alerting.
• LLM augmentation: Integrates Google's Gemini as an optional backend for natural language querying and deeper contextual analysis of environmental readings.
• User interfaces: Provides a web-based dashboard and Telegram bot integration for remote monitoring, alongside an LED strip for immediate visual status indication.

Why it matters

Atmosis demonstrates a practical template for integrating lightweight edge AI into low-cost open hardware sensor networks, showing how Edge Impulse on Arduino-class devices can move beyond raw data logging toward autonomous decision-making. The project's dual-interface approach, local web plus Telegram, offers a reproducible pattern for makers building distributed environmental monitoring without requiring dedicated mobile applications. For the Arduino ecosystem, winning challenge entries like this validate the UNO Q's dual-brain architecture as viable for real-world inference workloads and may influence future board designs targeting embedded machine learning.