Python Control System For Detection And Tracking Of Objects With Quadcopter Using Computer Vision

The research focused on the development of a control system in Python for the detection and tracking of objects using a low-cost quadcopter, based on artificial vision as a contribution in the educational field. The aim was to test the hypothesis that it is possible to implement these functions without resorting to expensive and proprietary drones. The methodology was divided into two stages: face detection and recognition, and quadcopter position control. In the first stage, the Haar cascade algorithm was used for face detection and the LBPH, FisherFaces and Eigenfaces models were evaluated for recognition. After a statistical analysis, the LBPH model was chosen for its effectiveness. The second phase consisted of a control system with a graphical interface (GUI) that allows the model to be trained, as well as to perform manual and automatic flights. For the automatic flight, a Derivative Proportional controller was used, suitable for real-time systems. Flight tests showed that the system can detect and track objects in low and high light conditions, achieving 96% accuracy at distances of 26 to 75 cm. It is concluded that the algorithm allows for facial tracking effectively, providing a solid basis for future research.