A.M. Andreev(1), V.S. Kolesnikov(2), A.V. Koriukin(3)

1,3 Bauman Moscow State Technical University (Moscow, Russia)

2 Penza State University (Penza, Russia)

1 arkandreev@gmail.com

Formulation of the problem. The relevance of this study is due to the growing interest in self-driving cars in recent years. One of the tasks of computer vision is recognition of road markings. This paper presents one of the ways to solve this problem on a prototype of an unmanned vehicle.

Goal. Implement an algorithm for recognizing road markings using the example of a prototype of an unmanned vehicle.

Results. The principle of operation of the Kenny operator and Hough transforms, which are used in the algorithm, are described. The minimum description of the prototype on which the algorithm was tested is given, and the model of the environment in which the testing was carried out is described. Each step of the algorithm is described, the images show intermediate results for each step. Based on the information obtained from the images, a method for calculating the angle of rotation of the wheels for the prototype is described.

Practical significance. One of the methods for solving the problem of detecting road markings is considered in here. Testing has shown that the algorithm works well on low-power devices such as Raspberry Pi 4 mini computer. This study contains popular methods for solving computer vision problems and will be useful to computer vision developers from various industries.

A.M. Andreev¹, V.S. Kolesnikov², A.V. Koriukin³

1,3 Bauman Moscow State Technical University (Moscow, Russia)

2 Penza State University (Penza, Russia)

1 arkandreev@gmail.com

1. «Vstraivaemye sistemy dlja avtonomnyh mashin novogo pokolenija». [Jelektronnyj resurs] URL: https://www.nvidia.com/ru-ru/autonomous-machines/embedded-systems. Data obrashhenija: 07.07.2020; (In Russian).
2. «Deep PiCar - Part 4: Autonomous Lane Navigation via OpenCV». [Jelektronnyj resurs] URL: https://toward­sdatascience.com/deeppicar-part-4-lane-following-via-opencv-737dd9e47c96. Data obrashhenija: 22.08.2020; (In Russian).
3. «Canny Edge Detection» [Jelektronnyj resurs] Data obrashhenija: 22.10.2020. Data publikacii: 23.03.2009. URL: http://www.cse.iitd.ernet.in/~pkalra/col783-2017/canny.pdf; (In Russian).
4. «Hough Line Transform» [Jelektronnyj resurs] Data obrashhenija: 27.10.2020. Data poslednej redakcii: 27.10.2020. https://docs.opencv.org/3.4/d9/db0/tutorial_hough_lines.html. (In Russian).