Bayesian Sensor Bias Correction for Location Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for determining the location and movement of computing devices, such as mobile phones, face inaccuracies due to sensor errors and the unavailability of GPS signals, leading to unreliable results over time.
Innovation Solution
The system employs a combination of sensors like gyroscopes, magnetometers, and accelerometers, using Bayesian filtering to calculate and remove sensor errors, iteratively determining sensor biases to enhance accuracy and mitigate errors, thereby improving location and movement determination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If sensor data is used to estimate location and movement, then location data can be obtained, but sensor errors cause inaccuracies that increase over time
Solution Approach 1:
The patent implements a feedback mechanism where sensor biases are continuously estimated and corrected using Bayesian filtering. The system uses GPS data when available to correct sensor measurements, creating a closed-loop feedback system that maintains accuracy over time by continuously adjusting sensor readings based on observed deviations from true values.
Solution Approach 2:
The patent dynamically changes sensor parameters by estimating and applying time-varying biases to sensor data. The system adjusts sensor readings by adding correction terms that compensate for drift and other errors, effectively transforming the sensor measurements to maintain accuracy despite changing environmental conditions and sensor degradation.
2Measurement precision
If GPS signals are used for location estimation, then accurate location data is obtained, but GPS signals may be unavailable in certain environments
Solution Approach 1:
The patent introduces sensor biases as intermediary correction terms that mediate between raw sensor data and accurate location estimates. These biases act as corrective intermediaries that bridge the gap between imperfect sensor measurements and true location values, enabling accurate estimation even when GPS signals are unavailable by continuously adjusting sensor readings based on observed patterns.
Solution Approach 2:
The system performs preliminary estimation of sensor biases using available GPS data during periods when GPS is accessible. These pre-calculated biases are then stored and applied during periods when GPS is unavailable, allowing the system to maintain accuracy by preparing correction data in advance during favorable conditions.
3Reliability
If multiple sensors are combined to improve location estimation, then measurement coverage is improved, but sensor errors and biases complicate the data processing
Solution Approach 1:
The patent segments the complex sensor fusion problem into separate correction processes for each sensor type. Instead of attempting to fuse all sensor data simultaneously, the system processes each sensor independently by estimating its specific biases and applying corrections, then combines the corrected readings. This segmentation reduces the complexity of the overall processing by breaking down the fusion task into manageable steps.
Data Source
AI summary
A system includes one or more processors, and data storage configured to store instructions that, when executed by the one or more processors, cause the system to perform functions. In this example, the functions include receiving sensor data that is collected by one or more sensors of a device over one or more locations and over a time period. Further, in the present example, the functions also include determining location estimates of the device by performing filtering of the sensor data to determine offsets for a least one sensor providing sensor data. The filtering is an iterative process of filtering control input data to determine the sensor bias based on data from a second sensor of the at least two sensors and adjusting the set of sensor data based on the determined bias.