EnviroScope is a portable personal weather and air-quality assistant designed to help people understand their indoor and outdoor environment in a practical and meaningful way. It combines multiple environmental sensors, a custom PCB, a 2.8-inch touchscreen interface, and an embedded machine learning model running directly on the device. Unlike regular weather stations that only display raw values, this system interprets the data and provides actionable recommendations—helping users improve comfort, health, and environmental awareness.

The device monitors PM2.5, carbon monoxide, temperature, pressure, humidity, UV, and ambient light indoors, and compares them with outdoor conditions. Everything—sensing, inference, and decision-making—happens at the edge.

Features

• Indoor & Outdoor Monitoring: Measures environmental conditions both inside and outside for complete situational awareness.
• Multi-Sensor Data Capture: Tracks PM2.5, CO, temperature, humidity, pressure, UV, and ambient light in real time.
• Edge Machine Learning: Runs an on-device ML model to analyze air quality and environmental patterns without cloud dependency.
• Actionable Recommendations: Converts sensor data into clear user guidance such as ventilation alerts, UV warnings, and comfort suggestions.
• Touchscreen Interface: 2.8-inch touch display for live readings, alerts, and intuitive navigation.
• Custom Hardware Design: Built on a compact, full custom PCB integrating sensors, microcontroller, and power management.
• Real-Time Alerts: Immediate visual notifications for unsafe or uncomfortable environmental conditions.

Motivation

I built this because most consumer weather or air-quality monitors stop at measurements. They show numbers, but they don’t tell users what those numbers mean, whether they should take action, or what action would help. Indoors especially, things like poor ventilation, elevated CO, or high humidity can affect comfort, productivity, and health, yet they often go unnoticed.

I wanted a device that could: sense the environment in detail, understand the context, and advise the user in real time.

I also wanted to challenge myself to integrate several disciplines—sensor fusion, embedded firmware, edge ML, hardware design, PCB layout, user interface, and power management—into a single working product. Training the ML model with Edge Impulse and deploying it gave me a complete “ML on embedded systems” workflow, which was both practical and educational. Building my own custom PCB made the design far more compact and product-like compared to a prototype on a breadboard.

Working

The system works in four major stages:

1. Sensing & Data Collection

Indoor sensors continuously measure:

• PM2.5
• CO2
• Humidity
• Temperature
• Pressure
• UV
• Ambient light

Outdoor conditions are also fetched from internet for comparison. The sensors feed data to a microcontroller on a custom PCB.

2. Local Processing & Edge ML

Raw sensor data is passed to an embedded machine learning model trained using Edge Impulse. The model performs real-time classification to determine environmental quality states (e.g., “good air,” “poor ventilation,” “high UV,” etc.). Running inference on-device means no cloud, no latency, and full privacy.

3. Decision-Making & Action Suggestions

Based on the model output and rule-based logic, the device generates suggestions such as:

“Ventilate the room”

“Avoid direct sunlight”

“Reduce heater usage”

“Humidity high—use dehumidifier”

This turns data into insight.

4. User Interface & Output

A 2.8-inch touchscreen shows:

• sensor readings
• indoor/outdoor comparison
• recommended actions
• warnings and alerts

The system is battery-powered and optimized for low power operation so users can place it anywhere, indoors or outdoors.

Conclusion

This project demonstrates a complete, end-to-end environmental monitoring and decision-assistance system that goes beyond traditional weather stations by transforming raw sensor data into meaningful, real-time guidance. By integrating multiple indoor and outdoor sensors, a custom PCB, a touchscreen user interface, and an edge machine-learning model trained with Edge Impulse, the device operates as a compact, battery-powered, and fully standalone solution. It not only measures environmental conditions but also interprets them to improve user comfort, safety, and awareness without relying on cloud connectivity. The project highlights strong interdisciplinary execution—combining embedded systems, hardware design, firmware development, and AI at the edge—while also presenting clear potential for future scalability into smart-home and consumer applications.