| Title |
Development of a Driver Route Guidance System using YOLOv8 |
| Authors |
노학기(Hak Ki Noh) ; 이석호(Seok Ho Lee) ; 정길도(Kil To Chong) ; 김지홍(Jee Hong Kim) |
| DOI |
https://doi.org/10.5573/ieie.2025.62.3.67 |
| Keywords |
Embedded system; Guide system; YOLO v8; Real time; Lane detection |
| Abstract |
In This research develops a real-time lane detection and guidance system by lightweighting YOLO v8 to improve the efficiency and quality of lane painting work. The NPU on the embedded board, khadas VIM4, is used to optimize the real-time inference performance of the lightweight YOLO v8 model and provide real-time lane marker and vehicle guidance information to the operator through a GUI. The hardware configuration consists of two Khadas VIM4s, two cameras, and a 485 communication module. The proposed system automates part of the painting process to reduce operator fatigue and ensure uniform lane painting quality. In our experiments, we achieved a performance of 22FPS for inference of the deep learning model and 12FPS for information provided by the GUI. This confirms that real-time information can be provided to a vehicle traveling at an average speed of 5km/h, enabling operators to accurately and efficiently perform painting tasks. |