What is Bias Instability of IMU Sensor?
IMUs are the key instruments of any modern navigation system, as they can be applied in autonomous driving, drones, robotics, etc.
Bias Instability is a highly significant parameter in the Inertial Measurement Units (IMUs) that describes minor variable variations of the measured sensor when it is outputting at zero, particularly in constant temperature and great environment. This is sometimes referred to as in-run bias stability, and beginning to emerge as an important contributor to a long term overall accuracy and reliability of an IMU.
The long term drift of sensor measurement is bias instability which occurs even when the IMU is stationary. Such a small but continuous error accumulates and causes huge errors in navigation.
Visualizing Bias Instability
Flicker Noise and White Noise:
What Matters about Bias Instability?
Due to bias instability, the position error increases with square of time in inertial navigation. To give an example, a gyroscope having a bias instability of 0.01°/hr will produce only 1 meter of position error in 1 hour, yet in 4 hours this will be 16 meters. This is an exponential increase, which renders bias instability a matter of concern in long-duration applications.
Bias Instability’s Measuring and Quantifying
1. Allan Deviation Analysis
Allan deviation analysis is the main technique that is used to measure the instability of bias. This method of statistics determines the various sources of the noise based on their variation with the averaging time.
2. Flat Region Identification
Bias instability in Allan deviation plot is represented by a flat area (smallest value). This minimum point is actually the optimal averaging time in which the bias instability is dominating.
3. Testing Conditions
Precise measurement needs controlled conditions: temperature stability, isolation of vibration and electromagnetic shielding. Tests are usually long several hours to record low-frequency drift.
4. Quantification
Bias instability (B) is determined at the minimum point of the Allan variance (σ²) where τ is the averaging time at the minimum: B = σ(τ)/√(2/π)
IMU Technologies Comparison
| Inertial Grade | Short-Term | Tactical/Self-driving/Humanoid robots | Navigation Grade |
|---|---|---|---|
| Example Applications | Flight Control | Missiles | Aircraft |
| Gyroscopes | >10 deg/hr | 0.1 – 10deg/hr | 0.001 – 0.1deg/hr |
| Accelerometers | >1mg | 100 μg – 1mg | 10 – 100μg |
Applications Affected by Bias Instability
Aircraft Navigation
To survive without a GPS, inertial navigation systems (INS) must exhibit ultra-low bias instability even in a long flight. In aviation grade IMU, bias stability is less than 0.01°/hr.
Autonomous Vehicles
When the GPS fails, self-driving cars use IMUs. The ability to steer properly following a tunnel or the canyons in the city depends on the instability of bias.
Augmented Reality
AR devices require a solid orientation tracking. When the bias instability is high, the virtual objects become drifting in relation to the real world, which terminates immersion.
Satellite Control
Maintaining Attitude control, Spacecraft must retain orientation over long periods of time, so attitude control systems must use gyroscopes with very low bias instability.
Mitigation Techniques
Sensor Calibration
Accurate calibration of the factories is a corrective approach to fixed bias. The temperature compensation takes into consideration thermal variation that influence bias stability.
Sensor Fusion
By fusing IMU data with GPS, magnetometers and even visual odometry with Kalman filters, bias instability keeps error growth under control.
Active Compensation
Advanced IMUs also have integrated heaters to keep the temperature constant and electrostatic actuators to keep recalibrating bias.
Material Science
New materials of MEMS such as silicon carbide and quartz are less thermally sensitive and vacuum packaging eliminates gas damping effects.
Key Takeaways
Bias instability is the major limitation to inertial navigation accuracy over time. Although the modern techniques have significantly enhanced the IMU performance, it is necessary to understand the bias instability:
- Choosing an appropriate IMU to suit your application in terms of duration and accuracy need
- Developing efficient algorithms of sensor fusion
- Forecasting performance of navigation systems in GPS denied environment
- Coming up with calibration methods that account to long-term driftt
With the development of autonomous systems and extended reality applications, instability of bias remains a key area of development of inertial sensors.