“With the rapid development of autonomous driving technology, urban street scene semantic segmentation technology, as a key link in autonomous driving technology, is gradually receiving attention. A new research achievement has made significant breakthroughs in the field of real-time urban street view semantic segmentation. The research team has proposed a real-time urban street view semantic segmentation algorithm based on a cross hierarchical aggregation network to address the challenges faced by current semantic segmentation algorithms, such as unclear pixel differentiation, inaccurate understanding of complex scenes, and inaccurate segmentation of small-scale objects. This algorithm achieves effective extraction and feature reuse of multi-scale contextual information by combining cross level aggregation pyramid pooling module, channel attention mechanism, and multi-scale fusion module, promoting the fusion of deep and shallow features. After verification on two universal urban street view datasets, Cityscapes and CamVid, this algorithm has achieved an effective balance between accuracy and real-time performance, and has significantly improved semantic segmentation performance compared to other algorithms. The experimental results show that the algorithm achieves an accuracy of 73.0% mIoU at 294 FPS on an RTX3090 graphics card, bringing new breakthroughs to the field of real-time urban street view semantic segmentation. This research achievement not only provides strong support for the development of autonomous driving technology, but also opens up new directions for research in the field of real-time urban street view semantic segmentation.”