HackStack

Project Info

TerraIntelligence

Devpost

Inspiration

Agriculture is one of the most data-rich yet data-scarce industries. A majority of farmers rely on intuition rather than data when making decisions about when, what, and how to farm. We wanted to build a system that brings real-time environmental intelligence to the field. The idea sprouted from how drones can capture so much stunning visual data, but weren't really used to collect other sensor data that are vital to driving decisions about agriculture.

What it does

TerraIntelligence is an AI-powered drone and IoT network for precision agriculture. It collects real-time environmental and geospatial data — including temperature, humidity, soil pH, water level, altitude, and distance — through onboard and ground sensors. That data is automatically streamed to a Firebase Realtime Database, visualized on a web dashboard, and used to map crop health trends or irrigation needs. The system can: Capture and tag images with precise location metadata (lat, lon, altitude) Stream live sensor readings to a web interface in real time Build a historical log of conditions across multiple zones Help identify patterns for water optimization, soil balance, or early signs of crop stress

How we built it

On the hardware layer, we used an ESP32 microcontroller connected to a DHT11 temperature-humidity sensor and ultrasonic water level module. The board connects to WiFi and pushes JSON-formatted data directly to Firebase RTDB. Additionally, we used a motor to launch and bring back up the payload (the ESP32 + sensor module) whenever it needed to collect data from a probing site. From there, the data is fetched by ML models designed to extract meaningful insights from the source data points. We used the NASA API and conducted literature review to decipher how this data can be used and why it’s meaningful for farmers. From there, we built a web dashboard that doesn’t just visualize data — it lets farmers actually interact with it. The dashboard fetches data from Firebase in real time, displaying it through live-updating charts and a leaflet.js map showing sensor locations and environmental readings. But we wanted to go beyond numbers and graphs, so we integrated AI-driven interaction. Farmers can: Download all their field data as CSV files for offline use or research. Talk to the dashboard directly using a VAPI voice agent, which allows natural, hands-free interaction: perfect for use out in the field. Ask questions about their data, powered by Claude, such as “Which area had the lowest moisture this week?” or “Show me pH trends for Zone 3.” The AI parses Firebase data and generates insightful, contextual responses on the spot.

Challenges we ran into

One of the biggest challenges was getting all of these different endpoints to work: we had data coming in from multiple sensors that had to be sent to the database, then analyzed by various ML models looking for various different signals, and finally sent to the frontend for users to interact with. We solved this through breaking down each component, and integrating together once each module worked as it needed to. Another major engineering problem was balancing the payload and ensuring that the drone would be able to support

Accomplishments we're proud of

We're proud of building the probe dropping mechanism on the drone, and all the data that it was able to collect. We also really liked how we were able to visualize and use multiple ML models to extract so much powerful data.

What's next

Getting farmers to actually use this. We have the mechanics, we have the software, we just need to find the people who need this.

Analysis

Compare with all teams

View

Metric

No commits on this project resolved to a GitHub account.

Technology

Found in codeClaimed only
  • AnthropicIn code
  • CSSIn code
  • FirebaseIn code
  • JavaScriptIn code
  • Next.jsIn code
  • ReactIn code
  • Tailwind CSSIn code
  • TypeScriptIn code

8 of 8 appear in the indexed code.

AI coding agents

  • Claude CodeCommits

Detected from committed agent config files and commit authorship. Absence of a signal is not proof an agent was unused.

Codebase size

Source size

1.2 MB

Source files

341

Counts recognized source files only; vendored directories, binaries and lockfiles are excluded, so this is smaller than the repository on disk.

0 stars