Droneventory

💡 Inspiration

Having worked in the restaurant industry as a bartender, I noticed firsthand how challenging inventory management can be. Restaurants experience high inventory turnover because food and drinks spoil under varying conditions, such as fluctuating demand and seasonal changes. This makes inventory prediction difficult, and manual counting is both time-consuming and often inaccurate. These challenges inspired me to create Droneventory, a solution to automate and streamline the inventory process, saving time and reducing errors.

🚀 What it does

Droneventory uses a Crazyflie drone equipped with object recognition to automate inventory counts. By scanning shelves and recognizing objects, it provides real-time updates, significantly reducing inventory errors and saving labor costs.

🛠️ How we built it

At the hackathon, I assembled the drone using a Crazyflie board integrated with an accelerometer, a pressure sensor, an ESP32 module, and a GAP8 microcontroller with 8 cores for running lightweight machine learning models efficiently. I connected it to Amazon Rekognition, creating an API gateway via AWS S3 to recognize objects through the camera integrated with the ESP32 over Wi-Fi.

🧩 Challenges we ran into

The biggest challenge was that the drone’s bootloader was outdated and didn’t support over-the-air flashing with the Crazyradio antenna. I needed a JTAG cable to flash my code, but none were available at the event. Unfortunately, my Amazon order for the cable is set to arrive after the hackathon.

🎉 Accomplishments that we're proud of

I'm proud to have made my first contribution to the open-source Crazyflie drone project, which is used by labs and research institutions worldwide. While my code hasn’t been accepted yet, submitting a pull request to such a significant project was a great milestone for me.

📚 What we learned

I learned a lot about the SPI communication protocol and how microcontrollers like the ESP32 and GAP8 work. I also gained experience in reading open-source projects with large codebases, identifying and modifying the right files for my contribution. This project deepened my understanding of how to interface with sensors and control hardware for drone navigation.

🔮 What's next for Droneventory

Next steps include finding a co-founder, refining the minimum viable product, and testing the system with local restaurants. Once proven, I plan to seek clients and scale the solution further.