Drone navigation system based on QR code recognition using fast component-based two-staged algorithm

Abstract

Kontes Robot Terbang Indonesia (KRTI) is a robotics competition organized by the National Achievement Center (Puspresnas). One of the branches of this competition is Vertical Takeoff and Landing (VTOL), which challenges participants to develop autonomous drones for package delivery based on QR code recognition. This research focuses on improving the accuracy of QR code recognition and drone navigation using a fast component-based two-stage method. Various image recognition methods for QR codes are discussed, including multi-threshold method, canny edge detection method, and fast componentbased two-stage method. The selected methods are modified to overcome the problems of rotation and perspective distortion. This research uses a quantitative and practical approach and is conducted indoors due to the limitations of drones on wind, light, and patterns on the ground. The fast component-based two-step method achieved 90.74% accuracy in QR code recognition, with certain challenges such as dark colored images. The navigation system relies on the drone’s position relative to the QR code in the image. The drone is instructed to approach the QR code for decoding. Test results showed 83.07% accuracy in matching commands, with problems due to QR code recognition failure. Strategies to overcome the recognition failure included rotation during takeoff and matching commands based on previous instructions. In addition, lag issues during drone navigation were addressed by optimizing the number of frames processed. The drone’s environmental requirements, such as well-lit areas and patterned floors, are described to ensure stable flight. This research provides insight into improving autonomous drone systems for package delivery through improved QR code recognition and navigation methods.