Adaptive ZRO Tracking Filter for Sensor Bias Compensation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current 3D pointing devices face challenges in accurately estimating and compensating for sensor bias, particularly in angular rate sensors, which leads to cursor drift and inefficiencies in motion tracking, especially during device motion and temperature changes.
Innovation Solution
An adaptive Zero-Rate Output (ZRO) tracking filter is implemented, combining a Kalman filter with cumulative and exponential moving-average filters, with dynamically modified gains and enforced constraints on predicted estimate covariance, to quickly converge to the true ZRO value and distinguish between human motion and stationary states.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional bias estimation methods are used in angular rate sensors, then the system is simpler to implement, but cursor drift occurs and motion tracking accuracy deteriorates during device motion and temperature changes
Solution Approach 1:
The patent implements a dynamic filter that adapts its behavior based on device state. The filter transitions between different modes (stationary vs. motion) and dynamically adjusts its parameters including the process noise covariance Q and measurement noise covariance R based on temperature changes and device acceleration detected by the accelerometer, allowing accurate bias tracking without excessive complexity
Solution Approach 2:
The patent introduces an accelerometer as an intermediary sensor to detect device motion and temperature changes. This intermediary provides additional information that helps the Kalman filter distinguish between device motion and cursor movement, improving bias estimation accuracy without requiring complex processing of the angular rate sensor data alone
2Measurement precision
If the ZRO filter converges quickly to compensate for sensor bias, then cursor drift is reduced, but the system becomes more sensitive to sensor calibration errors and motion detection errors
Solution Approach 1:
The patent implements beforehand cushioning by using the accelerometer to detect device motion and temperature changes before they significantly affect the bias estimation. The filter proactively adjusts its parameters in anticipation of changes, cushioning against the impact of calibration errors and motion detection errors on the final ZRO convergence accuracy
Solution Approach 2:
The patent implements feedback mechanisms where the Kalman filter continuously monitors the difference between predicted and actual sensor readings, and uses this feedback to adjust the bias estimate. The feedback loop incorporates temperature compensation and motion detection feedback to maintain reliability while achieving quick convergence
3Measurement precision
If the filter tracks ZRO during device motion, then motion tracking accuracy is improved, but the convergence time increases and computational complexity increases
Solution Approach 1:
The patent uses dynamic parameter adjustment based on device state. When motion is detected by the accelerometer, the filter switches to a motion-tracking mode with adjusted parameters that balance convergence speed and tracking accuracy. During stationary periods, the filter converges more aggressively, reducing overall convergence time while maintaining motion tracking accuracy when needed
Solution Approach 2:
The patent implements periodic evaluation of device state using the accelerometer to determine whether the device is in motion or stationary. This periodic state assessment allows the filter to switch between convergence modes, achieving fast convergence during stationary periods while maintaining accuracy during motion periods
Data Source
AI summary
A bias value associated with a sensor, e.g., a time-varying, non-zero value which is output from a sensor when it is motionless, is estimated using a ZRO-tracking filter which is a combination of a moving-average filter and a Kalman filter having at least one constraint enforced against at least one operating parameter of the Kalman filter. It achieves faster convergence on an estimated bias value and produces less estimate error after convergence. A resultant bias estimate may then be used to compensate the biased output of the sensor in, e.g., a 3D pointing device.