AR Pose Estimation Using Vision-Aided Sensor Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current vision-aided navigation systems for outdoor environments lack customization and accuracy in pose estimation, particularly when navigating arbitrary outdoor environments, as they do not effectively utilize environmental features like landmarks, horizons, and the Sun.
Innovation Solution
The system employs a combination of sensors, including cameras, accelerometers, gyroscopes, magnetometers, and GPS receivers, along with a processing module and database, using Extended Kalman Filter (EKF) algorithms and vision-aiding processes like landmark-matching, horizon-matching, and Sun-matching to enhance pose estimation by compensating for magnetic model bias and updating display pose vectors based on absolute azimuth measurements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional GPS/INS solutions are used for pose estimation, then basic navigation functionality is provided, but accuracy and customization for arbitrary outdoor environments are insufficient
Solution Approach 1:
The system continuously compares vision-based azimuth measurements with magnetometer-based predictions and uses the discrepancies to update and refine the pose estimation model, creating a feedback loop that improves accuracy over time
Solution Approach 2:
The system dynamically adjusts the weighting and processing of different sensor inputs (vision, magnetometer, GPS, IMU) based on environmental conditions and measurement quality to optimize pose estimation accuracy for specific environments
2Measurement precision
If vision-aiding processes are integrated to enhance pose estimation, then accuracy is improved, but system complexity increases
Solution Approach 1:
The patent combines multiple sensor systems (GPS, INS, vision, magnetometer) and processing algorithms (EKF, vision-aiding processes) into a unified pose estimation framework that shares common data structures and processing pipelines
Solution Approach 2:
The Extended Kalman Filter framework serves multiple functions: it processes data from diverse sensor types, handles different environmental conditions, and provides both pose estimation and attitude determination within a single unified system
Data Source
AI summary
The described technology regards an augmented reality system and method for estimating a position of a location of interest relative to the position and orientation of a display. Systems of the described technology include a plurality of sensors, a processing module or other computation means, and a database. Methods of the described technology use data from the sensor package useful to accurately generate signals to render graphical user interface information on a display, using vision-aiding processes, including horizon-matching, land-matching and Sun-matching.