Road Boundary Identification Using Laplacian Energy Features on Point-Cloud Images Generated from mmWave Radar

Accurate and real-time road boundary perception is essential for safe autonomous driving, especially when operating with lightweight sensors such as millimeter-wave (mmWave) radar. Existing radar-based boundary detection methods often rely on deep learning or computationally expensive fusion frameworks, which limit their suitability for real-time embedded deployment. In this paper, we propose a fast and efficient road boundary classification framework using a Laplacian Energy (LE)-based feature extraction method coupled with a Support Vector Machine (SVM) classifier. The proposed feature extraction technique generates compact discriminative descriptors, enabling highly efficient inference. Experimental results demonstrate that the proposed model achieves an accuracy of 87% while maintaining a prediction latency of only 16 ms, significantly outperforming traditional machine-learning baselines including Random Forest (RF), Logistic Regression (LR), Decision Tree (DT), XGBoost, and LightGBM.