Keeping tabs on plant and crop yield is a difficult job, even with the best farm equipment on hand. Some have gone the high-tech route using SBC-powered cameras to track growth and potential diseases. Still, researcher and developer Enrico Miglino has decided to use a customized drone for plant monitoring and data collection. His Nanodrone is semi-autonomous, and employs sensors and a camera to garner information in real-time along a predefined path using a GPS and a series of waypoints.
Miglino designed his Nanodrone using a DJ Mavic Mini and a 3D-printed undercarriage that houses an Arduino Nano 33 BLE Sense, a forward-facing 2MP SPI Arducam, a GPS board, and a microSD card (for saving information). As mentioned earlier, the Nanodrone uses a series of waypoints for navigation and the camera to identify colors (ripeness) of specific plants, such as fruits and vegetables, which it does leveraging TensorFlow Lite.
Once the Nanodrone has completed its navigation cycle, it then heads to a base station that’s equipped with a PSoC6 Pioneer Kit, where it then delivers the data it has collected via BLE. After a predetermined number of cycles (more than one may be needed), the collected data is then sent over Wi-Fi to a Raspberry Pi, which further processes the readings for more detailed information.
Miglino lists a number of possible applications for his Nanodrone, including plant and tree inspections on small and medium farms, architectural structure variations on time and deformation analysis, and environmental impact changes. For those who would like to recreate his build, Miglino has uploaded a detailed walkthrough of the Nanodrone on his element14 page as the project is still ongoing.