Common Issues of IMU Modules

Q: What are 6-axis, 9-axis, IMU, VRU and AHRS? The naming indicates the quantity and types of sensors integrated on the module.

• 6-axis: 3-axis accelerometer + 3-axis gyroscope (angular rate sensor)
• 9-axis: 6-axis plus 3-axis magnetometer

A 6-axis module can form VRU (Vertical Reference Unit) and IMU (Inertial Measurement Unit). A 9-axis module can form AHRS (Attitude Heading Reference System).

• IMU: Inertial Measurement Unit. It outputs acceleration and angular velocity without attitude angles.
• VRU: Built with IMU and embedded attitude solving algorithms to output attitude data.

When static, the accelerometer measures the gravity vector as a reference. Pitch and roll angles have low drift and high static precision. However, heading is perpendicular to gravity with no absolute reference, so heading error gradually increases over time.

When moving, accelerometers capture both gravity and motion acceleration (harmful acceleration). Gravity cannot be used to calibrate pitch and roll during motion. If the module stays under long-duration high-maneuver conditions, errors of all three Euler angles accumulate. Once back to static state, pitch and roll will recalibrate automatically, while heading remains uncorrected due to lack of reference.

A typical integrated navigation block diagram takes accelerometer, gyroscope, magnetometer, barometer and GNSS as inputs. Fusion algorithms output attitude, position and velocity.

Q2.1 Garbled or incorrect output data Check interface type: USB-TTL (3.3 V) and USB-232 cannot be interchanged. For example, CH110-232 outputs ±12 V 232 level and cannot connect to USB-TTL serial adapters. The baud rate must also match.

Q2.2 Wrong baud rate configuration Available baud rates: 9600, 115200, 256000, 460800, 921600, 1000000. Other values are invalid. If the baud rate is forgotten, test them one by one.

Q2.3 No quaternion output HI226/HI229 do not output quaternions by default. Quaternion output can be enabled via AT commands. Refer to the product manual.

Q3.1 What data can the module output? IMU consists of accelerometers and gyroscopes, so it directly outputs acceleration and angular velocity. The onboard processor calculates Euler angles (attitude) and linear velocity by integration. Linear velocity is usually not exported because integrated velocity error diverges rapidly. Position error becomes even larger after secondary integration. Some modules integrate magnetometers and barometers and can output their raw measurements.

Q3.2 Gyro alone can measure attitude, why use multi-sensor modules? Gyroscopes deliver good short-term attitude estimation by integrating angular rate, but inherent gyro errors exist. Accelerometers and magnetometers are added to stabilize and correct readings.

Q3.3 Output of static and horizontal placement When placed horizontally and static (ignoring Earth rotation): 3-axis acceleration output: [0,0,1 G]; 3-axis angular rate output: [0,0,0]. G stands for local gravitational acceleration.

Q3.4 Can the module integrate measurements to calculate velocity and position? Theoretically yes, practically not useful. Without external correction such as GNSS, position drifts quickly. For HI226, secondary integration may produce tens of meters of drift within one minute even when static; drift can reach 1 km under high-speed maneuvering. Stable pure inertial positioning relies on high-end IMUs such as fiber or laser gyroscopes, which are costly.

Q3.5 Susceptibility to strong magnetic interference from motors 6-axis modules are immune to magnetic interference. 9-axis modules suffer severe magnetic interference and are generally not suitable for robots or environments with magnetic objects.

Q3.6 Heading drift explanation Heading drift is inevitable for 6-axis modules and depends on sensor quality; algorithms cannot eliminate it completely. Heading is derived purely from gyro integration without absolute reference, leading to accumulated error over time. All attitude modules require roughly 1 second of static state after power-on to estimate gyro bias, otherwise heading drift worsens. 9-axis modules can output stable heading only under 9-axis mode, after magnetometer calibration and in environments free of magnetic distortion. Indoor environments such as offices, factories and laboratories have severe magnetic distortion, which degrades north-finding precision. Users are suggested to test 9-axis modules outdoors first.

Q3.7 Qualitative evaluation of dynamic precision without turntables This method only gives qualitative results: keep the module static and horizontal, then perform gentle random motion without exceeding gyro range for 2–5 minutes, then put it back horizontally. Pitch and roll will return to correct values. During movement, accelerometers lose gravity reference and attitude relies on gyro integration, accumulating errors. When static again, gravity provides reference and pitch/roll recover. The magnitude of recovery reflects gyro performance: larger recovery amplitude means larger accumulated gyro error. Upper computer software with curves or dashboards is required to observe the change.

Q3.8 Difference between quaternions and Euler angles Euler angles suffer from gimbal lock, which loses one rotational degree of freedom when two axes become parallel and restricts rotation into a degraded 2D space. Quaternions avoid gimbal lock and are recommended for complex rotation tracking. Euler angles are sufficient for simple tilt measurement.

Q3.9 Gyro sensitivity to vibration MEMS gyroscopes are highly sensitive to vibration. Although gyro theoretically measures angular velocity instead of acceleration, real MEMS devices have g-sensitivity (gravity sensitivity). In vibrating environments, g-sensitivity often matters more than bias stability. Low-cost MEMS IMUs perform far worse than fiber-optic gyros under vibration. Fiber-optic gyros have no such g-sensitivity due to different working principles. Mounting IMUs close to vibration sources such as fans will seriously degrade measurement precision.

Can RTK + INS + odometer realize full-scene all-weather positioning? Low-speed vehicle INS has poor error performance and reliability. Inertial navigation has cumulative error, and wheel odometer is affected by slipping and bumpy roads. Centimeter-level positioning for all scenarios remains challenging, comparable to the difficulty of autonomous driving. RTK suffers from signal occlusion and multipath effects under trees or tall buildings. Therefore LiDAR/camera mapping and high-definition map matching are widely researched. If full-scene all-weather positioning is achieved, autonomous driving can reach safety levels similar to trams and simplify regulation formulation.