Strong generalization ability, an inevitable choice for large-scale operation: Through continuous data iteration, it is easy to achieve rapid adaptation to different regions and application scenarios, greatly reducing algorithm development time and adaptation costs;

The only way out for large-scale commercial applications without high-precision map solutions is to no longer rely on high-precision maps, significantly improve deployment efficiency, and significantly reduce deployment and subsequent maintenance costs;

Realize data-driven closed-loop, model iteration automation: a complete and efficient data production pipeline, from raw data collection, automated labeling, model training, to OTA upgrades, a complete closed-loop toolchain and workflow, to achieve uninterrupted self iteration and performance improvement.