prann

a robot and iot based mesh for pollution anaylsis

Comment

Perfect — here’s a Hackathon project summary in the standard DevPost / Hackathon format for PRAANBOT-Cam, written in a clear, professional, and technically rich style suitable for submission or presentation 👇

🧠 Inspiration

The project is inspired by the severe air pollution crisis across major Indian cities, especially Delhi, where PM2.5 levels frequently exceed 300 µg/m³—nearly 10× the safe limit. Existing air-quality monitoring networks are stationary, expensive, and sparse, leaving large gaps in neighborhood-level pollution visibility.

We envisioned PRAANBOT-Cam — a mobile IoT robot that roams streets, captures real-time air data and live visuals, and uses AI models to predict and alert about upcoming pollution spikes. The name “Praan” (meaning life/breath in Sanskrit) symbolizes our goal: to protect every breath through intelligent technology.

⚙️ What It Does

PRAANBOT-Cam is an autonomous, low-cost robotic platform that:

  1. Monitors air quality using an MQ135 sensor (detecting CO₂, NOx, and smoke).
  2. Streams live visuals via ESP32-CAM for contextual pollution evidence.
  3. Navigates obstacles automatically using IR and ultrasonic sensors.
  4. Uploads data in JSON format to Baserow cloud database through Wi-Fi.
  5. Triggers Telegram alerts when dangerous pollution levels or anomalies are detected.
  6. Feeds data into Kaggle notebooks where AI models (LSTM, IsolationForest, K-Means) forecast pollution and identify hotspots.
  7. Visualizes everything on a live dashboard, showing video, AQ trends, and predictive analytics.

💡 In short: Robot patrols → Measures → Predicts → Alerts → Visualizes → Learns

🛠️ How We Built It

  1. Hardware Integration:
  • Main control via ESP32 DevKit V1 (for sensors and Wi-Fi).
  • ESP32-CAM streams video and transmits snapshots to the main ESP32.
  • Arduino UNO handles motor control (using L298N driver and DC motors).
  • Sensors: MQ135 (air), HC-SR04 (distance), IR (obstacle detection).
  • Custom power management circuit with Li-ion battery pack.
  1. Software & Networking:
  • ESP32 hosts a local Wi-Fi hotspot for internal communication.
  • Sends JSON telemetry to Baserow using its REST API.
  • Telegram Bot API connected to send automated AQI alerts.

  • Kaggle notebooks for data analysis:

    • LSTM → AQI Forecasting
    • IsolationForest → Anomaly Detection
    • K-Means + Folium → Hotspot Mapping
  1. Dashboard & Visualization:
  • ThingSpeak / Firebase Dashboard shows real-time air data, camera feed, and AI predictions.

🚧 Challenges We Ran Into

  • Synchronizing multiple microcontrollers (ESP32, ESP32-CAM, Arduino) with stable serial communication.
  • Handling Wi-Fi interference and bandwidth issues for live camera streaming.
  • Managing limited ESP32 memory while transmitting image data and JSON telemetry.
  • Power optimization for extended robot patrols using a small battery pack.
  • Designing a reliable obstacle-avoidance logic that works in outdoor, noisy environments.
  • Integrating AI inference pipelines (trained on Kaggle) with cloud data and real-time alerts.

🏆 Accomplishments That We’re Proud Of

  • Built a fully functional, mobile air-quality monitoring robot using low-cost components.
  • Achieved real-time AQI prediction accuracy of ~89% using the LSTM model.
  • Created a live alerting system via Telegram for high pollution or abnormal readings.
  • Established a Baserow-based data pipeline capable of logging, exporting, and visualizing telemetry continuously.
  • Demonstrated how AI + IoT + Robotics can come together to build climate-resilient technology.

📚 What We Learned

  • Deep understanding of IoT communication protocols (Wi-Fi, UART, REST APIs).
  • Practical experience in sensor calibration and data preprocessing for ML models.
  • Integration of real-time embedded systems with cloud AI workflows.
  • Value of data-driven environmental intelligence for sustainable urban planning.
  • How to manage distributed systems where hardware, software, and AI co-exist.

🚀 What’s Next for PRAANBOT

  1. GPS Integration: Enable geo-tagged pollution mapping and clustering.
  2. Swarm Network: Multiple PRAANBOT units forming a city-wide mesh for collective data.
  3. TinyML Models: Run prediction directly on the ESP32 for faster response (offline AI).
  4. Solar Charging: Improve autonomy and reduce downtime.
  5. Mobile App Interface: Citizen access to live AQI maps and pollution alerts.
  6. Partnerships: Collaborate with Delhi’s Smart City mission and environmental NGOs.

Long-term Goal: Create a scalable fleet of autonomous air-quality bots for real-time, street-level pollution analytics — empowering every community to act against air pollution.

Built With

Share this project:

Updates