Magnetic Sensor Calibration via Visual Inertial Odometry
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for calibrating magnetic sensors in client devices are inconvenient for users, as they require performing specific motions, such as a figure-eight pattern, to accurately calibrate the sensors.
Innovation Solution
The client device uses visual inertial odometry (VIO) data to determine its orientation and calibrate the magnetic sensor by computing a device correction vector, which accounts for the magnetic fields generated by the device itself, allowing for regular calibration without user intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing calibration methods using figure-eight motion patterns are used, then magnetic sensor calibration accuracy is improved, but user convenience deteriorates due to requiring repeated manual motions
Solution Approach 1:
The system performs magnetic sensor calibration automatically using visual inertial odometry data from the device's camera and motion sensors, eliminating the need for user-performed figure-eight motions. The device calibrates itself by processing video feed and inertial data to compute correction vectors, making the calibration process transparent and convenient for users.
Solution Approach 2:
The patent replaces the mechanical manual motion pattern (figure-eight physical movement) with an automated computational system using visual inertial odometry. The camera captures video frames and inertial sensors track device motion, which are then processed algorithmically to achieve calibration without requiring specific manual gestures.
2Measurement precision
If manual figure-eight calibration is performed, then magnetic sensor accuracy is improved, but calibration frequency is reduced due to user inconvenience
Solution Approach 1:
The device performs automatic calibration in the background using its camera and inertial sensors, enabling frequent recalibration without user intervention. The system continuously or periodically executes calibration routines, maintaining up-to-date correction vectors for magnetic sensor compensation.
Solution Approach 2:
The calibration process becomes a continuous or periodic background operation rather than a discrete manual task. The device maintains calibration by continuously processing visual and inertial data, ensuring magnetic sensor accuracy is preserved over time without requiring repeated user actions.
3Ease of operation
If automatic calibration using VIO data is implemented, then calibration frequency and convenience are improved, but device complexity increases due to additional processing requirements
Solution Approach 1:
The system leverages the device's existing camera and inertial sensors, which are already used for other functions like navigation and augmented reality. By repurposing this existing visual inertial odometry infrastructure for calibration, the patent avoids adding dedicated calibration hardware while achieving automatic calibration functionality.
Solution Approach 2:
The calibration process is merged with the device's existing visual inertial odometry processing pipeline. The same computer vision and inertial measurement systems used for other device functions are utilized for calibration, consolidating processing resources and reducing overall system complexity.
Data Source
AI summary
The disclosure describes a method for calibrating a magnetic sensor of a client device. The method may include receiving a set of magnetic field measurements, each of which includes a device location, an orientation of the client device, and an observed magnetic field vector measured by the magnetic sensor. The method may include computing a device correction vector for the client device based on the set of magnetic field measurements. For each magnetic field measurement, the method includes determining a world magnetic field vector at the device location of the magnetic field measurement, computing an expected measured magnetic field vector at the device location, accessing an estimated device correction vector for the client device, computing an expected adjusted vector for the client device, comparing the observed magnetic field vector associated with the magnetic field measurement and the expected adjusted vector, and computing the device correction vector based on the comparison.