Sensor Data Fusion for Accurate Navigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional inertial navigation systems using low-cost inertial measurement units (IMUs) suffer from significant numerical integration errors due to sensor noise, biases, and alignment errors, leading to inaccurate navigation data, especially in environments with limited GNSS coverage.
Innovation Solution
A method that processes multiple time periods of sensor data streams to combine measurements with motion models, analyzing data for reliability and accuracy by comparing instantaneous data with motion models and performing self-consistency checks to correct errors, thereby improving the accuracy of navigation data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If low-cost IMUs are used to reduce system cost, then device affordability increases, but measurement precision deteriorates due to sensor noise, biases, and alignment errors
Solution Approach 1:
The patent combines data from multiple sensors (accelerometers, gyroscopes, magnetometers, barometers, GNSS receivers) into an integrated sensor fusion system. By merging data from these different sensor types and processing them together through a unified navigation solution algorithm, the system compensates for individual sensor deficiencies and achieves accurate navigation data even with low-cost IMUs.
Solution Approach 2:
The patent implements a feedback mechanism where the navigation solution is continuously refined by comparing sensor measurements with expected values from motion models and previous navigation states. The system uses feedback from multiple time periods to identify and correct errors in navigation data, thereby improving measurement precision over time while maintaining device affordability.
2Productivity
If traditional single-epoch methods are used to process sensor data, then processing speed increases, but reliability deteriorates due to inability to assess data quality over time
Solution Approach 1:
The patent performs preliminary analysis of sensor data quality by examining multiple time periods before finalizing the navigation solution. The system pre-processes data to identify reliable measurements and detect potential errors or outliers, allowing it to weight or discard problematic data points before computing the final navigation result, thereby improving reliability without excessive processing overhead.
Solution Approach 2:
The patent employs dynamic processing where the system adapts its analysis depth and processing intensity based on data quality assessments. When data quality is high, processing can be more efficient; when quality is questionable, the system dynamically increases analysis depth by examining additional time periods, thus balancing productivity and reliability adaptively.
3Speed
If instantaneous data processing is used, then real-time performance is achieved, but measurement precision deteriorates due to inability to filter spurious data points
Solution Approach 1:
The patent applies partial filtering by selectively processing data from multiple time periods rather than all available data. The system identifies and processes only the necessary portions of historical data that contribute to filtering spurious points, applying excessive action only where needed to maintain real-time performance while improving measurement precision through targeted data validation.
Data Source
Figure 1
Figure 2~3
Figure 4~5
AI summary
A method of combining data from at least one sensor configured to make measurements from which position or movement may be determined. The method comprises: obtaining, in a first time period, first data from a first sensor mounted on a first platform (602); obtaining, in a second time period, second data from said first sensor and/or a second sensor mounted on the first platform or a second platform (603); determining the evolution of a metric of interest between first and second time instances, wherein at least one of the first and second time instances is within at least one of the first and second time periods, and wherein the evolution of the metric of interest is constrained by at least one of the first data, second data and a motion model of the first and/or second platform (604).