This study introduces an innovative method for detecting potholes utilizing Light Detection and Ranging (LiDAR) technology combined with machine learning algorithms. Due to the unavailability of public LiDAR datasets dedicated to pothole detection, synthetic data is generated to create a dataset suitable for training and validating machine learning models. The methodology centers on converting 3D LiDAR point cloud data into 2D height maps, from which essential features—such as surface roughness, gradient metrics, and depth variations—are extracted. Two machine learning classifiers, Support Vector Machine (SVM) and Random Forest, are employed for binary classification of road surface conditions. Model performance is assessed using metrics such as accuracy, precision, recall, and F1-score. Additionally, feature importance analysis and the characterization of pothole depth are conducted to gain deeper insights. The findings indicate that LiDAR-based techniques can effectively identify road surface irregularities while maintaining computational efficiency.