Inspiration
A few weeks ago, my dad's tomato garden got infested by pests. To save the plants, he had to spray pesticides, but shortly after, we found a bunch of dead bees scattered around the garden. It made me realize that farmers and gardeners are often forced to spray completely in the dark, with no way of knowing if beneficial insects are active right above their crops. After researching how unmonitored the airspace above fields really is, we realized agriculture is in desperate need for a sort of weather forecast for insects, and that inspired PollenPath.
What it does
PollenPath tracks pollinator activity above crop fields and gives farmers real-time data so they don't accidentally kill beneficial bugs. A sky-facing sensor sits right at crop canopy height, using an upward infrared beam to detect the wingbeat frequencies of flying insects. It classifies them into bees, hoverflies, moths, and butterflies without using cameras. The application automatically tells farmers the safest 2-hour windows to spray, locking out sprays when winds top 15 km/h, during temperature inversions, or when harmless pollinators are out in high numbers.
How we built it
We used Gemini and Claude to help brainstorm, refine the detection math, and write code for the web platform. The app was built using Google AI Studio. We modeled the optical sensor head, solar panel mount, and radiation shield in CAD, and built the physical demonstration prototype by hand using cardboard, paint, and craft materials on a strict deadline.
Challenges we ran into
Our biggest challenge was physically assembling the hardware prototype under a tight time crunch with limited supplies. Translating our CAD models into a clean, accurate physical cardboard build while simultaneously building and testing the companion web app pushed our team to the wire.
Accomplishments that we're proud of
We are really proud of the sheer number of practical features we packed into the project. Instead of just showing raw sensor numbers, we built a complete end-to-end system: optical wingbeat detection physics, pesticide drift and weather inversion safety logic, hive deficit alerts, and an easy-to-use map dashboard for farmers.
What we learned
We learned just how crucial time management is during a fast-paced hackathon. On the technical and environmental side, we learned a ton about aerosphere ecology, how temperature inversions trap pesticide clouds, how wind drift works, and how much agriculture relies on invisible insect highways flying overhead.
What's next for Untitled
Our goal is to take PollenPath from a hackathon concept to a real product farmers can buy and install. We'll replace the cardboard prototype with durable, 3D-printed, weather-sealed enclosures with real silicon photodiodes and microcontrollers, train machine learning models on live insect flight data to improve classification accuracy, and partner with local farms to test the stations in real orchards and fields.
Log in or sign up for Devpost to join the conversation.