A Deep Learning Framework on Embedded ADAS Platform for Lane and Road Detection

Modern vehicles are now equipped with the lane detection, a critical safety feature that alerts drivers when their vehicle veers off the road [1]. ResNet-18 excels in lane detection, leveraging a 18-layer architecture with residual learning to overcome gradient challenges in deep networks. It plays a crucial role in driver assistance systems and autonomous vehicles by accurately identifying and marking lanes on roads. The network’s depth and hierarchical feature learning empower it to distinguish subtle lane markings from surrounding elements in complex road scenes. Here the Hardware-in-the-Loop (HIL) simulation is used for testing and validation, the LDWS system, integrated with a lane detection deep learning framework and Nvidia Jetson Xavier AGX, can be evaluated across diverse conditions, including varied road geometries, climatic scenarios, and traffic patterns. This approach provides a realistic and reliable platform for comprehensive testing in a virtual environment, facilitated dSPACE Scalexio. Our algorithm achieved an accuracy of 94% at daylight and an accuracy of 81% at night [2].