Vehicle Heading Determination Using Magnetometer Feature Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining a vehicle's heading, particularly in low GPS signal conditions or when the vehicle is stationary, are inaccurate due to reliance on GPS bearing data, which can lead to suboptimal route planning and reduced travel efficiency.
Innovation Solution
A method and system that utilize a magnetometer to determine the vehicle's heading by extracting features from magnetic flux density data, generating a feature vector, and using a trained classifier, such as a random forest algorithm, to predict the heading, even when the vehicle is stationary or moving slowly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If GPS bearing data is used to determine vehicle heading, then the method is simple to implement, but the accuracy deteriorates when GPS signal is poor or vehicle is stationary
Solution Approach 1:
The patent segments the heading determination process into multiple independent feature extraction components from magnetometer data (e.g., horizontal component, vertical component, declination angle, inclination angle). Each feature is processed separately and then integrated, allowing the system to maintain accuracy across different vehicle states (stationary, slow-moving, fast-moving) while keeping the overall system manageable in complexity
Solution Approach 2:
The patent introduces magnetometer data as an intermediary measurement source between GPS and the final heading determination. When GPS bearing is unreliable (poor signal or stationary vehicle), the system switches to using magnetometer-derived features as the primary heading source, effectively mediating the transition between different measurement reliability states
2Measurement precision
If magnetometer data with multiple features is extracted and processed through a classifier, then the heading determination accuracy is improved, but the computational complexity increases
Solution Approach 1:
The patent performs preliminary action by pre-training a classification model (e.g., random forest, support vector machine, or neural network) offline with labeled heading data. The model learns optimal feature combinations and decision boundaries in advance, so that during runtime, the system only needs to extract features from magnetometer data and feed them into the pre-trained model, significantly reducing real-time computational complexity while maintaining high accuracy
Solution Approach 2:
The patent transforms raw magnetometer data into multiple derived parameters (horizontal component, vertical component, declination angle, inclination angle, and their combinations). This parameter transformation enriches the feature space, allowing the classifier to make more accurate heading determinations. The system dynamically adjusts which parameters are used based on vehicle motion state, optimizing the balance between accuracy and computational load
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach provides accurate heading determination in various conditions, improving route planning efficiency and reducing time wastage for both drivers and passengers by leveraging machine learning algorithms to process magnetometer data effectively.
Implementation Method 1
retrieving sensor data generated by a magnetometer of the mobile computing device within one or more first time slots; determining, based on the sensor data, a plurality of magnetic flux densities along three axes of a Cartesian coordinate system
Data Source
AI summary
A method and system for determining a heading of a vehicle. The method may include receiving, from a mobile computing device, a request for determining the heading of the vehicle. The method may also include retrieving, at least in response to the request, sensor data generated by a magnetometer of the mobile computing device within one or more first time slots, and obtaining a classifier trained to determine a predicted heading of the vehicle. The method may further include, for the sensor data of each of the one or more first time slots, obtaining a feature vector by extracting features from the sensor data, and determining, based on the obtained feature vector, a predicted heading by inputting the feature vector into the classifier. The method may also include determining the heading of the vehicle based on the obtained one or more predicted headings.