IMU Sensor Fusion Using Machine Learning for Navigation Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Inertial navigation systems face challenges in accurately fusing measurements from multiple inertial sensors due to differences in performance characteristics and environmental factors, leading to accumulated navigation errors that are difficult to characterize and compensate for using traditional methods.
Innovation Solution
The use of machine learning algorithms to create models for extracting features, selecting sensors, applying weights, and compensating measurements from multiple inertial sensors, enabling dynamic weighing and automation of the modeling process, and reducing the effects of noise and errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional fusion methods are used to combine measurements from multiple inertial sensors, then the system structure remains simple, but navigation errors accumulate and accuracy deteriorates
Solution Approach 1:
The patent introduces machine learning algorithms as an intermediary between raw sensor measurements and navigation solutions. The fusion model acts as a mediator that automatically processes measurements from multiple inertial sensors, extracting features and applying optimal fusion strategies without requiring complex manual intervention or simplified traditional fusion methods.
Solution Approach 2:
The fusion model performs self-service by automatically learning optimal fusion strategies from data. The machine learning algorithm autonomously identifies patterns, extracts relevant features, and adjusts fusion parameters without external intervention, enabling the system to adapt to different sensor characteristics and environmental conditions automatically.
2Extent of automation
If manual feature extraction and weight application are used in sensor fusion, then the system remains interpretable, but the process is time-consuming and automation level is low
Solution Approach 1:
The patent replaces manual mechanical processes of feature extraction and weight application with machine learning algorithms. The automated fusion model substitutes human-operated traditional fusion methods with computational algorithms that automatically process sensor data, extract features, and determine optimal weighting strategies in real-time.
Solution Approach 2:
The machine learning-based fusion model enables continuous automated processing of inertial sensor measurements. Unlike manual methods that require periodic intervention, the automated system continuously extracts features, applies weights, and generates navigation solutions without interruption, maintaining optimal fusion performance throughout operation.
3Adaptability or versatility
If traditional fusion methods are applied to measurements from sensors with different performance characteristics, then the system structure remains simple, but the ability to compensate for environmental factors and sensor errors is insufficient
Solution Approach 1:
The fusion model applies local quality by treating each inertial sensor individually based on its specific performance characteristics. The machine learning algorithm analyzes and adapts to the unique properties of each sensor, applying localized fusion strategies that account for individual sensor errors, noise levels, and environmental sensitivities rather than applying uniform fusion methods to all sensors.
Solution Approach 2:
The patent implements parameter changes by dynamically adjusting fusion parameters based on environmental conditions and sensor performance. The machine learning model modifies weighting factors, fusion strategies, and compensation parameters in real-time according to changing conditions, enabling the system to adapt to different environmental factors and maintain reliability across varying operational contexts.
Data Source
AI summary
Systems and methods for multiple inertial measurement unit sensor fusion using machine learning are provided herein. In certain embodiments, a system includes inertial sensors that produce inertial measurements, a memory unit that stores a fusion model produced by at least one machine learning algorithm, and a processor that receives inertial measurements, where the processor applies the fusion model to the inertial measurements. The fusion model directs the processor to extract features from the inertial measurements, and to select inertial measurements based on a sensor in the plurality of inertial sensors that produced the inertial measurements. Also, the fusion model directs the processor to apply weights to the selected inertial measurements based on the extracted features, to apply compensation coefficients to the selected inertial measurements, and to fuse the selected inertial measurements into an inertial navigation solution.