Author | Intelligent Driving Community
Breakdown of the Draft for "Technical Requirements and Test Methods for Vehicle Positioning Systems Part 2: Inertial Measurement Unit"
In tunnels, underground garages, and under overpasses, once satellite signals are lost, how does the vehicle know where it is? The answer is the Inertial Measurement Unit (IMU). It does not rely on any external signals but "senses" motion through its own gyroscopes and accelerometers, relying entirely on it to maintain positioning continuity in the short term.
However, there was previously no national automotive standard defining what constitutes qualified IMU performance. In April 2026, the draft for "Technical Requirements and Test Methods for Vehicle Positioning Systems Part 2: Inertial Measurement Unit" was released, marking the first time the technical requirements and test methods for automotive IMUs were incorporated into a national standard draft.
This article breaks down this draft from start to finish: what it specifies, what exactly each metric restricts, how the tests are designed, and what it means for engineers.
01 What is This Standard: The "Second Part" of the Positioning System Trilogy
This standard is Part 2 of GB/T "Technical Requirements and Test Methods for Vehicle Positioning Systems". The entire series is divided into three parts: Part 1 Satellite Positioning, Part 2 Inertial Measurement Unit, and Part 3 Fusion System. Part 1 has already been published, Part 3 on fusion positioning is also being drafted, and Part 2 is currently in the public consultation stage.
The standard was proposed by the Ministry of Industry and Information Technology and is under the jurisdiction of the National Technical Committee of Auto Standardization (SAC/TC114). The drafting units cover OEMs, Tier 1 suppliers, and positioning manufacturers such as NUAA, CATARC, Asensing, Bosch, BYD, Changan, Desay SV, and Qianxun Spatial Intelligence—basically, the rules were set jointly by the upstream and downstream of the industry chain.
The scope of application is clearly defined: it applies to automotive IMUs containing 3-axis accelerometers and 3-axis gyroscopes, while other inertial measurement units may refer to it for implementation. In other words, the standard targets the most common 6-axis IMU configuration of "gyroscope + accelerometer".
02 Output Requirements First: The "Minimum Configuration" for Autonomous Driving Systems
Clause 5.1 of the standard first specifies the output requirements, which are the prerequisite for all performance metrics. The output of the automotive IMU must include three types of information: 3-axis acceleration, 3-axis angular velocity, and temperature. Temperature information is listed as a separate item because the bias and scale factor of MEMS devices are extremely sensitive to temperature, and outputting temperature is the foundation for subsequent compensation.
Here is a critical point: for IMUs applied in driving automation systems, the output update rate should not be lower than 100 Hz. Why 100 Hz? Because the control cycle of autonomous driving systems is typically in the order of 10 milliseconds, an insufficient IMU update rate would cause a break in the data chain for dead reckoning and sensor fusion filtering. For IMUs applied in the cockpit or other functions, the update rate can be negotiated between the supplier and the buyer—the standard does not adopt a one-size-fits-all approach, leaving room for engineering flexibility.
03 Nine Metrics for Gyroscopes: What Exactly Does Each One Restrict
Performance requirements are the core of the standard. For IMUs targeting driving automation systems, gyroscopes must pass 9 metrics, and accelerometers must pass 8 metrics. Let's look at the gyroscope first.
Bias ≤ 0.2 °/s: Bias is the average angular velocity output by the gyroscope when stationary, equivalent to the illusion of "rotating while at rest". The limit of 0.2 °/s corresponds to an acceptable angular velocity drift in engineering; exceeding this magnitude would make the static reference for attitude estimation unstable.
Bias Stability ≤ 10 °/h: Bias stability measures the dispersion of the bias over time, directly affecting the drift rate of the heading angle over time. 10 °/h means the heading error is controllable after long-term operation, which is a common threshold for automotive navigation-grade IMUs.
Bias Instability ≤ 4 °/h: This metric is calculated using the Allan variance method, reflecting the lower limit of random bias fluctuations over a long time scale. It is usually regarded as the "noise floor" metric for MEMS gyroscopes and is also a core parameter for distinguishing device grades.
Angle Random Walk ≤ 0.35 °/√h: Angle random walk describes the angular error accumulated over time after the integration of white noise. Together with bias instability, it determines the upper limit of accuracy for short-term dead reckoning.
The remaining metrics are equally critical: Full-temperature bias stability ≤ 0.1 °/s, assessing bias drift under temperature variations; Scale factor nonlinearity ≤ 2×10⁻⁴, assessing input-output linearity; Full-temperature scale factor error ≤ 3‰, assessing the stability of the scale factor across the full temperature range; Cross-axis coupling ≤ 0.05°, assessing inter-axis crosstalk; Bandwidth ≥ 80 Hz, ensuring the response capability to dynamic inputs.
Figure 1: Quick Overview of Core Metrics for Automotive IMUs (Schematic diagram, compiled based on the draft for comments)
This figure summarizes the 9 metrics for gyroscopes and 8 metrics for accelerometers. It can be seen that the standard assesses "room-temperature static performance" and "full-temperature dynamic performance" separately—passing at room temperature does not mean passing at full temperature, which is precisely the most essential difference between automotive and consumer-grade scenarios.
04 Eight Metrics for Accelerometers: Stricter Accuracy Requirements
Accelerometer bias ≤ 5 mg, bias stability ≤ 0.1 mg, bias instability ≤ 50 μg, velocity random walk ≤ 0.04 m/s/√h. Putting these numbers together, it is evident that the standard's requirements for accelerometers are quite high.
In particular, the bias instability of 50 μg is already several times higher than the requirements for consumer-grade devices. The reason lies in the fact that the accelerometer bias is directly integrated into velocity error, and the velocity error is then integrated into position error, resulting in a double integration amplification. In tunnel scenarios where GNSS is invalid for a long time, the performance of the accelerometer directly determines the rate of positioning drift.
Scale factor nonlinearity ≤ 3×10⁻⁴, full-temperature scale factor error ≤ 3‰, and cross-axis coupling ≤ 0.05°—these three metrics are at the same magnitude as those for gyroscopes. Cross-axis coupling assesses the crosstalk between sensitive axes; installation errors and process deviations can both cause coupling. The stricter the metric, the higher the requirements for packaging and calibration processes.
05 Environmental Tests: A Harder Hurdle Than Performance
IMU performance metrics are measured under laboratory conditions, but once installed in a vehicle, it must face the real automotive environment. Section 5.3 of the standard divides environmental requirements into five categories: electrical performance, electromagnetic compatibility (EMC), mechanical performance, environmental weatherability, and chemical load.
Electrical performance is executed according to the GB/T 28046 series, covering DC supply voltage, overvoltage, superimposed AC voltage, slow decrease and increase of supply voltage, transient variations, reverse voltage, reference ground and supply offset, open circuit, short circuit protection, and insulation resistance. The insulation resistance is required to be greater than 10 MΩ, which is the baseline for safety.
EMC requirements mandate that conducted and radiated emissions meet GB/T 18655-2025 Class 3. Immunity tests cover electrical transients on power lines (pulses 1, 2a, 2b, 3a, 3b, Class III), capacitive and inductive coupling, electromagnetic radiation immunity, and electrostatic discharge. In autonomous driving systems, IMUs are often located in the same area as radars, cameras, and antennas; failing EMC means becoming a solid source of interference or a victim.
Mechanical performance includes mechanical vibration, mechanical shock, and free fall. Environmental weatherability covers low-temperature storage and operation, high-temperature storage and operation, temperature gradient, temperature cycling, rapid temperature changes, damp heat cyclic, steady-state damp heat, solar radiation, and dust and water protection. Operating temperature range: -40 ℃ to 85 ℃ when there are no special requirements, and -40 ℃ to 90 ℃ when installed in direct sunlight within the passenger compartment or on the roof.
Protection ratings are divided into two types: IP5K2 for the cabin interior and IP6K9K for the cabin exterior. Exterior installation means facing direct rain washing and high-pressure spraying. IP6K9K is the protection rating against high-pressure and high-temperature water jets, which poses a significant test for the structural design and potting process of the IMU.
Chemical load tests are executed according to GB/T 28046.5, assessing resistance to oil, solvents, cleaning agents, etc. After the test, the surface should not blister, crack, peel, or corrode, and markings and labels should remain clear.
06 Test Methods: Turntables, Thermal Chambers, and Allan Variance
Chapter 6 of the standard provides complete test methods, which are worth a close look by engineers, because whether the metrics can be measured and whether the measured results are trustworthy entirely depend on the test design.
The requirements for test equipment are quite specific: the orthogonality error of the turntable axis system should not be greater than ±3″, the angular position positioning error should not be greater than ±3″ (dual-axis), and the rate range should not be less than ±300 °/s; the thermal chamber temperature range is -55 ℃ to 105 ℃, with a loaded temperature change rate of not less than 3 ℃/min; the angular vibration table frequency range is 0.1 Hz to 150 Hz, with a maximum angular acceleration of not less than 15000 °/s². These equipment specifications themselves represent the threshold for high-precision testing equipment.
Performance tests are divided into six categories: room-temperature static, room-temperature multi-position, room-temperature multi-rate, full-temperature static, full-temperature dynamic, and bandwidth testing. Room-temperature static testing requires data recording for at least 4 hours—both bias stability and Allan variance analysis require long-term data support, and short-term sampling cannot measure true noise characteristics.
Full-temperature static testing follows a cycle of "reference temperature - minimum temperature - maximum temperature - reference temperature" for at least 3 cycles, with a holding time of no less than 1 hour for each temperature change stage. Full-temperature dynamic testing is conducted at 7 temperature points (-40, -20, 0, 25, 45, 65, 85/90 ℃), and at each temperature point, the scale factor error is tested using forward and reverse rotation at ±60 °/s.
The most core aspect of data processing is Allan variance analysis: through the Allan variance curve under double logarithmic coordinates, least squares fitting is used to obtain various coefficients such as quantization noise, random walk, bias instability, and rate random walk, and then bias instability and angle/velocity random walk are calculated. This method originates from the field of precision instruments and is now written into automotive standards—indicating that the assessment of automotive IMUs is approaching the rigor of navigation-grade devices.
07 Inspection Rules and Engineering Implications
Inspection rules group samples into 7 sets: Group 1 for output and performance tests, Group 2 for environmental weatherability and insulation resistance, Group 3 for mechanical vibration, Group 4 for dust and water protection and chemical load, Group 5 for EMC and solar radiation, Group 6 for mechanical shock and free fall, and Group 7 for durability (optional). All type test items must be qualified; if one item is unqualified, double sampling can be conducted for re-inspection.
The durability test uses the Arrhenius model to calculate the test duration. Appendix A of the standard provides the complete algorithm: activation energy EA=0.45 eV, combined with the temperature distribution model of the installation location to calculate the acceleration factor. In the example, for an IMU installed in direct sunlight within the passenger compartment, calculated based on 6000 hours of operating time, the high-temperature durability test duration is approximately 1156 hours.
For engineers, this standard provides at least four implications. First, tiered performance metrics: IMUs for autonomous driving systems must meet all performance requirements, while other applications such as the cockpit can be negotiated, making the criteria clearer during procurement and technology selection. Second, full-temperature assessment is a hard threshold: good room-temperature metrics do not mean automotive qualification; full-temperature bias stability and full-temperature scale factor error directly determine whether a device can pass certification. Third, there is a threshold for test equipment: turntables with ±3″ accuracy and angular vibration tables from 0.1 to 150 Hz represent new investments for testing institutions and supplier laboratories. Fourth, Allan variance becomes a mandatory test item: the data processing methods for random walk and bias instability are standardized, and the discrepancy between the "nominal values" and "measured values" of devices will be eliminated.
It should also be noted that this is currently a draft for comments, with the standard number still using the GB/T XXXXX placeholder. Metrics and test details may be adjusted before the official release. The discussions during the feedback period are precisely the window for engineers to participate in rule-making.
Satellite positioning gives the vehicle "absolute coordinates", the IMU gives the vehicle "motion intuition", and the fusion system stitches the two together. This IMU standard first establishes the baseline for "motion intuition".
Source:
Technical Requirements and Test Methods for Vehicle Positioning Systems Part 2: Inertial Measurement Unit, Plan No. 20250898-T-339 | Automotive Standard Formulation and Revision Management System of China Automotive Technology and Research Center