Inertial Navigation Filter Recovery from Invalid GNSS Observations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Inertial navigation systems (INS) face reduced accuracy and potential filter divergence when relying on inaccurate observations from satellite-based positioning systems, such as GNSS, leading to biased navigation solutions and prolonged periods of inaccurate navigation.
Innovation Solution
An apparatus and method for error mitigation in INS that involves storing historical filter parameters and sensor samples in a history buffer, validating observations before updating filter parameters, and processing sensor samples without immediate updates based on potentially inaccurate reference system observations to recover a navigation solution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If filter parameters are updated based on GNSS observations, then navigation solution accuracy is improved, but filter divergence occurs when observations are inaccurate
Solution Approach 1:
The system performs preliminary validation of GNSS observations before updating filter parameters. A validation module checks observation quality metrics (signal-to-noise ratio, geometric dilution of precision, residual errors) in advance to determine whether observations are reliable. Only validated observations trigger filter parameter updates, preventing inaccurate data from causing filter divergence while maintaining accuracy improvement from valid observations.
Solution Approach 2:
A validation module acts as an intermediary between the GNSS observation receiver and the filter parameter updater. This intermediary layer evaluates observation quality and selectively permits or blocks parameter updates based on validation results. The intermediary ensures that only high-quality observations reach the filter update mechanism, resolving the contradiction between utilizing observations for accuracy and preventing divergence from inaccurate ones.
2Measurement precision
If filter parameters are updated continuously with GNSS observations, then navigation accuracy is maintained, but system complexity increases due to validation requirements
Solution Approach 1:
The validation module utilizes inherently available GNSS signal characteristics (signal-to-noise ratio, geometric dilution of precision, carrier phase measurements) to perform self-validation without requiring external reference systems or additional sensors. The system validates its own observations using built-in GNSS measurement data, maintaining continuous accuracy improvement while avoiding the complexity of external validation infrastructure.
3Reliability
If filter restarts are performed to correct divergence, then filter stability is restored, but navigation solution continuity is interrupted
Solution Approach 1:
The system applies preliminary anti-action by validating observations before they can cause filter divergence in the first place. By checking observation quality metrics and blocking potentially harmful updates beforehand, the system prevents divergence rather than correcting it afterward. This eliminates the need for filter restarts and maintains continuous navigation solutions, resolving the contradiction between restoring stability and maintaining continuity.
Data Source
AI summary
A method, apparatus and computer program product are configured to mitigate error in an inertial navigation system (INS) that relies upon a reference system to determine or update filter parameters. In a method, a determination is made of (i) deviation or divergence of a filter of the INS and/or (ii) invalidity of observation(s) by the reference system. The method also identifies the filter parameters from a history buffer associated with a prior point in time at which performance of the filter is stable. The filter parameters were updated at least once based upon observation(s) by the reference system after the prior point in time. The method forward processes sensor samples of the INS obtained subsequent to the prior point in time using the filter parameters from the history buffer at which performance of the filter was stable to mitigate the error.