During the build phase of our drone prototype, our team integrated the core flight and AI systems into a lightweight, functional design. A key change was optimizing the drone body by switching to a 3D-printed carbon nylon frame to reduce weight while supporting all components, including four motors, a stack (containing both a flight controller and an ESC), and a battery.
We mounted a Raspberry Pi 4B paired with a Pi AI camera for real-time object detection, using a ribbon strip and integrated Walk Snail GPS for navigation and positioning. We also had to tweak power distribution and component layout to maintain flight balance and stability.
This phase was critical in translating our concept into a working prototype, allowing us to test flight dynamics, component integration, and onboard AI functionality. It laid the foundation for refining detection accuracy, stability, and future software enhancements.
Code of AI object detection feature and the window that pops up outlining the objects that the AI camera can see
Syncing up the drone and its motors with the Ardupilot software to ensure the drone meets the initial testing conditions
Pairing up the controller with the drone to be able to control the drone's flight and maneuver it
Unsoldering some extra ports of the Pi 4 to reduce weight of the Pi and make sure the drone doesn't have too heavy of a load to carry
The v1 drone was shaking because of the instability of the frame
The v2 design of the drone frame
v2 drone is able to fly!