HD Map

HD Map (High Definition Map) is a high-precision, high-dimensional digital map specifically engineered for autonomous vehicles. Unlike conventional navigation maps designed for human drivers that merely outline basic road network topological connections, HD maps achieve centimetre-level absolute spatial accuracy and incorporate comprehensive 3D geometric data, semantic layers, and intricate traffic regulations. It serves as critical core infrastructure for Advanced Driver Assistance Systems (ADAS) and high-level autonomous driving (such as Level 3 and above) to execute lane-level localisation, beyond-line-of-sight perception, smooth trajectory planning, and fail-safe redundancy decision-making.

Core Technical Architecture and Data Dimensions

The technical architecture of an HD Map fundamentally comprises multi-layered high-precision data models, a cloud-based dynamic closed-loop update mechanism, and an on-vehicle computing and matching engine:

Multi-Layer Semantic Data & Centimetre-Level Geometric Topology:

Structurally, HD maps are categorised into multiple layers: the road network layer (defining road topology and lane connectivity), geometry layer (comprising centimetre-level lane centre lines, road curvature, and gradients), semantic layer (precisely detailing traffic signages, physical coordinates of traffic lights, road markings, road edges, guardrails, median barriers, and 3D landmarks such as high-voltage pylons), and dynamic layer (delivering real-time traffic congestion and roadwork updates). 

Multi-Source Crowdsourced Data Collection & Cloud-Based Real-Time Freshness Closed Loop:

Generating and maintaining HD maps initially requires dedicated survey vehicles equipped with LiDAR and high-precision inertial navigation systems (RTK-GPS) for comprehensive real-world environment scanning. In the mass-deployment phase, real-time road condition changes are crowdsourced via mass-produced vehicles equipped with on-board cameras and millimetre-wave radars. Through cloud-based big data analytics and automated AI filtering, modifications such as road diversions and newly erected road signs are updated efficiently, ensuring the ongoing "freshness" of map data.

Safety Protocols and Physical Limitations

Strict Prohibition of Blind Reliance on HD Maps for High-Level Autonomous Driving During Roadworks, Adverse Weather Obscuring Road Markings, or Critical Map Data Outages: 

Despite their immense precision, real-world road conditions remain inherently dynamic. Blindly engaging lane-level autonomous driving is strictly prohibited when encountering extensive snow-covered roads, sudden unmapped road diversions, or blind spots where satellite signals fail to sync with HD maps. Under such conditions, the system must rely on on-board "perception-heavy" capabilities, or the driver must take over manual control immediately. 

Strict Adherence to Cross-Border Surveying Regulations and Regional Compliance Licensing for HD Maps: 

As HD maps house vast volumes of high-precision geospatial coordinates and 3D infrastructure data, they directly impact national geospatial security. Importing uncertified foreign or cross-regional HD map data lacking regulatory surveying credentials and official compliance into local vehicle systems is strictly prohibited. Automakers and end-users must strictly adhere to local surveying and mapping laws as well as data security compliance standards.