🤖 AI Summary
Autonomous driving maps face concurrent challenges in achieving high precision, lightweight representation, and end-to-end integration. To address these, this paper proposes a three-stage evolutionary paradigm: High-Definition (HD) maps, Lite maps, and Implicit maps—systematically analyzing their representational forms, production pipelines, and fundamental bottlenecks. We introduce the first unified taxonomy integrating semantic compression, neural radiance fields (NeRF), differentiable rendering, and end-to-end learning to enable a paradigm shift from explicit geometric storage to implicit neural representation. Our contributions include a full-lifecycle technical roadmap, a multi-stage collaborative mapping framework, and an automated production pipeline. These advances significantly improve map generalizability and deployment efficiency. The work establishes both theoretical foundations and practical guidelines for next-generation autonomous driving map research and development.
📝 Abstract
Maps have always been an essential component of autonomous driving. With the advancement of autonomous driving technology, both the representation and production process of maps have evolved substantially. The article categorizes the evolution of maps into three stages: High-Definition (HD) maps, Lightweight (Lite) maps, and Implicit maps. For each stage, we provide a comprehensive review of the map production workflow, with highlighting technical challenges involved and summarizing relevant solutions proposed by the academic community. Furthermore, we discuss cutting-edge research advances in map representations and explore how these innovations can be integrated into end-to-end autonomous driving frameworks.