What is IMU Noise Density?
Inertial Measurement Units (IMUs) are electronic devices which measure and report on the precise force, angular rate and orientation. Noise Density is one of the most important specifications which represents the innate chaotic variations that the IMU’s output produces when it’s not moving. Understanding this parameter is important in order to choose the correct IMU for your application and to predict how it will perform.
Noise Density is usually measured in units of μg/√Hz in the case of accelerometers, or °/s/√Hz or rad/s/√Hz for gyros. It is the noise power per unit bandwidth and is useful to engineers for determining how the noise will integrate over time to cause position and orientation errors.
Key Concepts of Noise Density
Power Spectral Density
Noise density is the square root of what is known as a Power Spectral Density (PSD), which is a way of describing the distribution of the power of a signal by frequency.
White Noise Characteristic
IMU noise is usually “white” in the majority of its bandwidth, i.e. has equal power per frequency interval.
Angle Random Walk
For gyroscopes the density of noise directly affects the Angle Random Walk (ARW) which affects orientation accuracy.
Velocity Random Walk
For accelerometers, the noise density is the determining factor for Velocity Random Walk (VRW), which makes a difference in the accuracy of the position by position through double integration.
Impact of Noise Density on IMU Performance
1. Short-term vs Long-term Effects
The high frequency performance is the area that noise density causes the most problem. For short term measurements, the lower the noise density the less jitter is in the measurement.
2. Integration Drift
When integrating angular rate to obtain angle or acceleration to obtain velocity and position, noise accumulates over time, producing drift that affects the length of time that an IMU can be used without external correction.
3. Bandwidth Considerations
The total noise power is the noise density times the square root of the bandwidth. Thus, constraining bandwidth can help to reduce the noise, but at the expense of response speed.
4. Sensor Fusion Impact
In Kalman filters and other sensor fusion algorithms, the noise density is used to fix the process noise covariance and this directly influences the performance and stability of the filter.
Noise Density in Different Applications
Consumer Electronics
With cost and power limitations overwhelming extreme precision requirements, IMUs with moderate noise density (100-500 μg/√Hz for accelerometers) are used in smartphones and wearables.
Automotive
Vehicle navigation systems need medium performance Inertial Measurement Units (IMUs) with (50-100 μg/√Hz) to handle stability control, navigation and advanced driver assistance systems.
Aerospace & Defense
Aircraft and missile guidance systems require ultra-low noise IMUs (10-50 μg/√Hz) for navigation accuracy without frequent updates from GPS systems.
Industrial & Robotics
Industrial navigation and robotic control systems are generally required to be low noise IMUs (25-75 μg/√Hz) for accurate position and control of motion.
Noise Density Comparison
| IMU Grade | Accel Noise Density (μg/√Hz) | Gyro Noise Density (°/hr/√Hz) | Typical Applications |
|---|---|---|---|
| Consumer | 100-500 | 0.01-0.05 | Smartphones, gaming devices |
| Industrial | 50-100 | 0.005-0.01 | Robotics, platform stabilization |
| Tactical | 25-50 | 0.002-0.005 | Autonomous vehicles, drones |
| Navigation | 10-25 | 0.0005-0.002 | Aerospace, marine navigation |
| Strategic | <10 | <0.0005 | Missile guidance, submarine navigation |
Reducing Noise Effects in Practice
Filtering Techniques
Proper filtering can help to greatly reduce the noise without stated price reduction to performance. Kalman filters, complementary filters and low-pass filters are commonly used.
Sensor Fusion
Combining IMU data with GPS, magnetometers, or sighting systems can be used to correct for drift caused by noise and achieve better overall accuracy.
Calibration
Temperature calibration, compensation can be used to reduce the effect of lesions to prevent the impact on noise measurement caused by environmental factors.
Oversampling
Sampling at a higher rate than required and then averaging can work well to reduce noise (by increasing resolution of measurement).
Conclusion
Noise density is one of the basic specifications that directly affects the performance of the IMU, especially in some applications that require high precision of navigation or motion tracking. By understanding this parameter and its effect on your system you can make informed decisions on how to choose a particular IMU, how to design the system and choose a signal processing approach. Remember that the lower the noise density the better the performance usually is, but then often the higher the cost and power consumption is, so balance your requirements against these.