IMAGE PROCESSING TECHNOLOGY AND DEEP LEARNING MODELS FOR DEVELOPING SELF-DRIVING CARS
Keywords:
Self-driving car, Lane detection, Object detection, Robot Operating System (ROS), Exponential Moving Average (EMA)Abstract
This paper presents the methodology and development plan for a 1/10-scale self- driving car operating in a miniature city. This includes simulation implementation, path planning, and data collection for lane detection and traffic signal recognition using image processing and deep learning techniques. The Robot Operating System (ROS) is the fundamental platform for developing robotic and autonomous vehicle systems. ROS facilitates communication between individual software components through topics and actions. The Gazebo simulation software in ROS is used to build and configure maps, vehicles, and objects to create simulations replicating real-world environments. Lane detection is performed using image processing, edge detection, and the Hough Transform algorithm to identify lanes, calculate the steering angle, and p2025rovide predictions for vehicle navigation. Object and traffic signal data are collected, labeled, and trained using the YOLOv5 model. The detected data are then used to generate control signals, which are transmitted via ROS and serial communication to a control board for motor and servo actuation.
References
Li, Y., & Chen, L. (2016). Nighttime lane markings recognition based on Canny detection and Hough transform. In Proceedings of The 2016 IEEE International Conference on Real-Time Computing and Robotics.
DOI: 10.1109/RCAR.2016.7784064
Lee, S., Son, H., & Min, K. (2010). Implementation of Lane Detection System using Optimized Hough Transform Circuit. In Proceedings of the IEEE International Conference. Korea Electronics Technology Institute, Seongnam-si, Republic of Korea. ISBN: 978-1-4244-7456-1.
DOI: 10.1109/APCCAS.2010.5775078
Pannu, G. S., Ansari, M. D., & Gupta, P. (2015). Design and Implementation of Autonomous Car using Raspberry Pi. International Journal of Computer Applications, 113(9), March 2015.
DOI: 10.5120/19854-1789
Tran Ngoc, H., & Quach, L. D. (2022). Adaptive Lane Keeping Assist for an Autonomous Vehicle based on Steering Fuzzy-PID Control in ROS. International Journal of Advanced Computer Science and Applications (IJACSA), 13(10).
DOI: 10.14569/IJACSA.2022.0131086
Roy, S., & Zhang, Z. (2020). Route Planning for Automatic Indoor Driving of Smart Cars. In Proceedings of the 2020 IEEE 7th International Conference on Industrial Engineering and Applications. Tongji University, Shanghai, China.
DOI: 10.1109/ICIEA49774.2020.9102061
Hwang, K., Jung, I. H., & Lee, J. M. (2022). Implementation of Autonomous Driving on RC-CAR with Raspberry PI and AI Server. Webology, 19(1), January 2022.
DOI: 10.14704/WEB/V19I1/WEB19293.
Hồ Văn Thu, & Lê Thanh Phúc. (2015). Ứng dụng xử lý ảnh nhận dạng đường đi cho ô tô chạy tự động (Application of Image Processing to Detect Lane for Autonomous Vehicle). Tạp chí Khoa học Giáo dục Kỹ thuật, Trường Đại học Sư phạm Kỹ thuật TP.HCM, (31/2015), 21.
https://jte.edu.vn/index.php/jte/article/view/492
Li L, Wang Z, Zhang T. GBH-YOLOv5: Ghost Convolution with BottleneckCSP and Tiny Target Prediction Head Incorporating YOLOv5 for PV Panel Defect Detection. Electronics. 2023; 12(3):561.
DOI: 10.3390/electronics12030561
Xie, J., Pang, Y., Nie, J., Cao, J., & Han, J. (2022). Latent feature pyramid network for object detection. IEEE Transactions on Multimedia, 25, 2153-2163.
DOI: 10.1109/TMM.2022.3143707
Wang, H., Qin, Q., Chen, L., Li, Y., & Cai, Y. (2025). RTMDet-R: A Robust Instance Segmentation Network for Complex Traffic Scenarios. IEEE Transactions on Intelligent Transportation Systems.
DOI: 10.1109/TITS.2025.3539658
