This paper introduces two models, RoadBEV-mono and RoadBEV-stereo, for road elevation reconstruction in Bird’s Eye View (BEV) to improve the performance of autonomous vehicles. The models leverage monocular and stereo images, respectively, to estimate road elevation with high accuracy. RoadBEV-mono directly fits elevation values based on voxel features queried from the image, while RoadBEV-stereo recognizes elevation patterns based on the BEV volume representing the correlation between left and right voxel features. The paper demonstrates the effectiveness and superiority of these models through experiments on a real-world dataset, achieving elevation errors of 1.83 cm and 0.50 cm for RoadBEV-mono and RoadBEV-stereo, respectively. The models are promising for practical road preview, enhancing safety and comfort in autonomous driving. The code is available at <https://github.com/ztsrxh/RoadBEV>.This paper introduces two models, RoadBEV-mono and RoadBEV-stereo, for road elevation reconstruction in Bird’s Eye View (BEV) to improve the performance of autonomous vehicles. The models leverage monocular and stereo images, respectively, to estimate road elevation with high accuracy. RoadBEV-mono directly fits elevation values based on voxel features queried from the image, while RoadBEV-stereo recognizes elevation patterns based on the BEV volume representing the correlation between left and right voxel features. The paper demonstrates the effectiveness and superiority of these models through experiments on a real-world dataset, achieving elevation errors of 1.83 cm and 0.50 cm for RoadBEV-mono and RoadBEV-stereo, respectively. The models are promising for practical road preview, enhancing safety and comfort in autonomous driving. The code is available at <https://github.com/ztsrxh/RoadBEV>.