Self-owned data ingestion vehicles developed in-house to ensure data precision

Building data crowd-sourcing ecosystem to ensure data freshness

Processing data rapidly with the tool chain developed in-house to enable rapid imaging

Building mature integrated business systems to ensure data quality

Based on the landmarks of positioning map layer, current vehicle location can be calculated accurately by matching with the sensing results and calculating the distance, which is redundant with the high-precision positioning results obtained from other technical approaches to ensure the vehicle is aware of its own position all the time.

HD maps provide information like precise road surface domain and surrounding facilities, and mark objects like sidewalks as areas of interest. As the objects in the areas of interest might influence the driving behavior of the autonomous driving system, it’s necessary that the autonomous driving vehicle uses precise model to detect the objects in the areas of interest, and calculate such information as type, speed and posture of each object.

Lane-level route planning and local route planning are major applications of HD maps. The global route planning result from starting point to the destination is synchronized to the autonomous driving system, matching with HD maps and referring to the result to achieve lane-level route planning, the result of which is smooth, safe local route planning.

HD maps accurately record the correlations between lanes, such as the passing rules and traffic rules. In autonomous driving scenarios like complicated crossings with flows of vehicles and pedestrians, they can help the vehicle significantly improve the precision of prediction about the behavior of every road-rights competitors to make optimal decisions.