Automotive High-precision Positioning Research Report, 2025

High-precision positioning research: IMU develops towards ""domain controller integration"" and ""software/hardware integrated service integration""

According to ResearchInChina, in 2024, the penetration rate of advanced autonomous driving (L2.5 highway NOA, L2.9 urban+highway NOA) in China’s domestic passenger cars reached 11.4%, and this figure is projected to rise to 31.6% by 2030. In highway NOA scenarios, lane-level/decimeter-level positioning is sufficient to identify the vehicle’s lane. However, in urban NOA scenarios, due to narrow lanes and complex road conditions, centimeter-level positioning accuracy is required.

As the cornerstone of autonomous driving perception, high-precision positioning holds vast market potential and demands continuous technological advancement.

Future Directions for High-Precision Positioning Development

Direction 1: IMU Integration into Domain Control and Integrated Software/Hardware/Service in End-to-end autonomous driving solution

In end-to-end autonomous driving systems, integrating IMU into domain controllers has become mainstream. With the evolution of autonomous driving, vehicles require higher accuracy and safety in perception and positioning, especially for L3+ autonomous vehicles that demand ultra-high-precision positioning.
Integrating the high-precision positioning module into the autonomous driving domain controller via SMD can reduce data transmission, effectively shorten information delay, and improve positioning accuracy. Specifically, there are several types of architectural design patterns:
IMU integrated into autonomous driving domain controller, with GNSS placed in T-BOX.
Autonomous driving domain controllers integrating both GNSS chips and IMU modules.

IMU and GNSS combined into an INS module integrated into domain controller.

DJI’s integrated ""LiDAR + Vision + IMU"" solution, with GNSS chips embedded in autonomous driving domain controller.

In addition, there are still a series of engineering problems that need to be solved when integrating the IMU module into autonomous driving domain controller:
Time synchronization: High-level intelligent driving vehicles have rigid requirements for the real-time performance and accuracy of positioning data. At the same time, poor real-time positioning will reduce positioning accuracy;
Functional safety: Because the intelligent driving domain controller integrates many functional modules, the system is relatively complex, and it usually has a multi-core and multi-process operating environment and complex application data flow, which makes it more difficult to achieve functional safety. OEMs need a functional safety solution with a complete RTK&INS combined algorithm deployed on the domain controller;
Temperature compensation: When integrating the IMU, which is very sensitive to temperature changes, with the GNSS and domain controller, it is necessary to consider the damage that the operating temperature may cause to the product.

In order to help the product application of OEMs, Sixents Technology launched a ""Smart Driving Software, Hardware and Service Integrated Positioning Terminal” in 2024H1. As an organism that is flexibly combined with multiple product lines, the terminal can not only realize the functions of the original P-BOX, but also acts as a priori platform for high-precision positioning domain controller integrated solutions. The high-precision positioning capability has shifted from P-BOX to domain controllers, which requires all product lines to be broken down. This terminal can greatly accelerate the promotion of integrated solutions. In the future, more and more smart driving software, hardware and service integrated positioning solutions will be implemented in various domain controllers.

Sixents Technology's ""Smart Driving Software, Hardware and Service Integrated Positioning Terminal” has a complete closed loop with high-precision positioning capabilities. Its four product baselines - GNSS (satellite navigation), IMU (inertial navigation), PE (terminal algorithm), and CS (differential correction service) have been owned and productized by Sixents Technology after years of R&D and verification. Among them, CS, as the first commercialized product line, has been mass-produced for passenger cars, commercial vehicles, industry applications and other fields. PE has been involved in multiple mass-production projects. Newer GNSS and IMU product lines have been included in mass-production projects.

Sixents Technology’s “Smart Driving Software, Hardware and Service Integrated Positioning Terminal” is the first to integrate GNSS+IMU+PE+CS, reducing costs by 30% compared to traditional solutions.

In addition, based on Sixents Technology's globally integrated high-precision positioning services, positioning solutions at home and abroad can be unified, and OEMs do not need to replace software and hardware solutions when going overseas. Sixents Technology’s ""Smart Driving Software, Hardware and Service Integrated Positioning Terminal”, based on global differentiated services and terminal algorithm groups, can fully cater to a unified architecture in the global market and system-wide localization, forming a stable, safe, and reliable system closed loop.

Sixents Technology has offered over 10 billion times of high-precision positioning services per day, with more than 3,000 sites, more than 15 million designated vehicles of over 40 models.

Direction 2: Multi-sensor fusion positioning achieves SLAM (simultaneous localization and mapping)

The IMU is built into the camera module and uses the visual-inertial navigation system (VINS) to use acceleration data to improve the accuracy and stability of binocular ranging. High-precision vehicle trajectory calculations can be maintained in some extreme environments, such as emergency traffic jams, heavy rain, nighttime, underground parking lots, etc.

Relying on the technological accumulation of UAV systems, DJI began to explore an inertial navigation stereoscopic binocular vision system in the field of intelligent driving as early as 2016. After years of precipitation, the system has matured. In 2025, an inertial navigation trinocular and LiDAR assembly system will be launched, with the cost much lower than the current “LiDAR + vision + P-Box” solution.

In addition to DJI, Sixents Technology has also developed a visual perception positioning terminal that integrates technologies such as binocular cameras, IMU, GNSS, RTK, and visual fusion positioning algorithms.

Direction 3: Integration of GNSS and INS, as well as integration of GNSS chip and IMU module

In order to adapt to intelligent driving system integration and cost reduction, Tier1 suppliers have launched GNSS-INS integrated solutions.
BYNAV Technology's SMD-type automotive-grade GNSS/INS high-precision integrated navigation module can be integrated into the domain control and smart driving solutions of Tier1 and other vendors to improve the overall integration of smart driving systems

Aceinna's SMD-type high-precision positioning module design solution supports the integration of INS modules composed of IMUs and GNSS into domain controllers

Sixents Technology’s GNSS and INS Positioning Terminal: GNSS and IMU products will be rapidly iterated to enable a closed-loop positioning system solution ecosystem.

Direction 4: MEMS IMU chips realize integrated calculation and integrate AI algorithms

STMicroelectronics launched the ASM330LHBG1 automotive-grade AI inertial module to improve vehicle navigation and positioning accuracy and reliability
STMicroelectronics has launched the ASM330LHBG1 automotive-grade inertial module, which integrates a three-axis MEMS accelerometer, a three-axis MEMS gyroscope module and a safety software library. Equipped with ST’s machine-learning core (MLC) and programmable finite state machine (FSM), the ASM330LHBG1 can run artificial-intelligence (AI) algorithms in the sensor to provide smart functionality at low power.

Asensing Technology’s GST80 programmable MEMS IMU chip with built-in MCU

At the Beijing International Automotive Exhibition in April 2024, Asensing Technology unveiled its new MEMS IMU chip - GST80. This chip is the industry's first programmable automotive sensor with a built-in MCU, featuring high integration, high performance and cost-effectiveness. GST80 is completely independently designed, developed, packaged, tested and produced by Asensing Technology, and the entire production process is conducted domestically, effectively ensuring the security of the supply chain and filling the domestic gap in high-performance automotive-grade MEMS IMU chips.

High-precision positioning from the perspective of OEMs: lower hardware costs and higher algorithm performance

OEMs always seek to reduce costs and increase efficiency. On the one hand, IMU integration is achieved through powerful domain controllers to reduce hardware costs, and SD pro MAP is introduced to replace expensive HD MAP to cut down map costs. On the other hand, the complexity of map algorithms has been greatly improved, such as model-based IMU noise reduction, online calibration and temperature compensation, and the combination of vision, LiDAR and IMU for high-precision map positioning.

As autonomous driving technology continues to mature, more and more vehicles will use laser SLAM and visual SLAM technology to achieve SLAM. The tight coupling of LiDAR, vision and IMU can improve the accuracy, reliability and robustness of positioning and attitude estimation, and is particularly suitable for scenarios with high-speed dynamics, complex environments, many occlusions or unreliable GPS.

For example, XPILOT 4.0 has improved the positioning accuracy from the previous decimeter level to the centimeter level, and transferred the ""GPS-based positioning"" to ""vision and IMU-based positioning"".

In order to handle urban positioning, Xpeng has chosen a solution of signal-independent vision + IMU positioning, which is not bound by GPS signals but is extremely difficult to achieve visual positioning.

Leapmotor has upgraded the fusion positioning architecture, using the original IMU and GNSS, visual semantics and radar to perceive the environment, and then combining HD/SD maps to achieve an overall fusion positioning framework.


Please Note: PDF E-mail from Publisher purchase option allows up to 10 users and does not allow printing or editing. This functionality will require a Global Site License.