Guangzhou Asensing Technology Co., Ltd. demonstrated HD-MapBox, a mapping application that integrates high-precision map data based on high-precision positioning, at CES 2022. The application can achieve lane-level positioning and 1+ mile (2km) Predictive Cruise Control (PCC), providing a decision basis for advanced assisted driving to better meet the demanding positioning requirements of autonomous driving vehicles.
Asensing Technology CTO Situ Chunhui said, “As the premise for autonomous driving safety, high-precision positioning is of great importance for integrating positioning technology based on inertial measurement units (IMU), global navigation satellite system (GNSS) signals, visual perception system and HD map. High-precision positioning is becoming the preferred choice due to higher positioning accuracy and improved redundancy as well as an enhanced passing rate under all scenarios.”
Empowering a highly reliable autonomous driving system
Under any driving scenario, autonomous driving vehicles must accurately interpret their own lane-level location information to better predict and prevent risks and make safe driving decisions. As a result, positioning is not only part of the autonomous driving process but also the premise of autonomous driving.
However, any single positioning technology has its own limitations, especially in certain scenarios such as in tunnels and underground garages where the perception system may be adversely affected by changes in the amount of light and low GPS signal, thereby affecting driving safety.
Based on data fusion of the GNSS, IMU, ADAS camera, vehicle dynamics and HD-map, the new HD-MapBox launched by Asensing Technology is a superior solution for lane-level position. With this robust collection of data on hand, the HD-MapBox can achieve a lateral error of less than 8 inches (0.2 meters) and a longitudinal error of less than 6.5 feet (2 meters) with a 95 percent confidence interval, providing an accurate reference for highway pilot (HWP) and automated valet parking (AVP). Even if both GNSS and lane line detection are not available, the HD-MapBox can still enable vehicles to keep in lane for at least a quarter mile (400 meters).
Taking AVP with its relatively high positioning requirements as an example, the scenarios where the driverless parking solution is most applicable are mainly in underground parking facilities, especially within facilities that require navigating complex, twisting turns along a ramp or passing through areas with poor lighting. These various scenarios present several risks in actual application as ordinary ADAS cameras may not operate as normally, and the satellite positioning function may also fail.
After incorporating high-precision sensor and map data, these deficiencies are fully compensated for through accurate mapping of the driving trajectory and parking space location with dead reckoning (DR) and camera parking space detection in the construction stage of the parking space map. When automated valet parking is in progress, the vehicle is guided to a parking space or driven to the assigned parking space through the integration of several positioning technologies.
Asensing Technology founder and CEO Rongxi Li said, “With the further development of advanced assisted driving in recent years, high-precision integrated positioning technology will be more widely used, especially in terms of improving driving safety and optimizing the intelligent driving experience. High-precision integrated navigation has become a basic configuration in more and more autonomous driving systems.”