Gökmen-PG: Path Recognition and Tracking in Wild Forests Using Deep Learning and OpenCV
In this video, we demonstrate the autonomous capabilities of the Gökmen-PG project, specifically designed for path recognition and tracking in wild forests without the use of GPS. The video displays the testing of our computer vision algorithms and controller code, operating autonomously using the SITL technique in the Gazebo simulation environment.
Gökmen-PG employs a Convolutional Neural Network model (PGNet) and a combination of OpenCV's segmentation functions to autonomously identify and follow trails in challenging unstructured environments. This video presents a side-by-side comparison of these two techniques, demonstrating how deep learning methodologies have been effectively employed to imbue Gökmen-PG with its unique autonomous path-finding ability.
We plan to further advance this autonomy by testing the SITL software in real-time autonomous flights soon. Stay tuned for further updates on the progress of the autonomous Gökmen-PG project.
For more details:
gokmenuavteam.com
instagram.com/gokmenuavteam