Spatial Feature Localization for XR Pose Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Inaccurate localization of electronic devices in XR environments leads to incorrect placement of virtual content, as changes in the physical environment can disrupt the estimated pose of the device, causing misalignment of XR content.
Innovation Solution
The method involves obtaining an environmental model with spatial feature locations, determining the device's pose based on these features, and updating the pose estimate by comparing actual spatial feature locations detected in the environment with the model, using visual inertial odometry and depth sensors to refine the localization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If visual inertial odometry is used to estimate device pose, then the device can obtain location estimates without additional sensors, but the pose estimation becomes inaccurate when physical environment changes occur
Solution Approach 1:
The system continuously compares detected spatial features against the environmental model and uses the discrepancies to generate feedback signals that correct the pose estimate. This feedback loop enables the system to maintain accurate localization even when visual-inertial odometry becomes inaccurate due to environmental changes.
Solution Approach 2:
The patent replaces the mechanical/sensor-based visual-inertial odometry system with a computational geometry-based approach using spatial feature detection and environmental modeling. This substitution allows the system to achieve higher accuracy without requiring additional physical sensors.
2Measurement precision
If the environmental model is updated frequently to reflect changes, then the localization accuracy improves, but the processing time and computational load increase
Solution Approach 1:
Instead of updating the entire environmental model frequently, the system performs partial updates only when and where necessary based on detected feature changes. This selective updating approach maintains localization accuracy while significantly reducing computational overhead compared to full model updates.
Solution Approach 2:
The environmental model is pre-built and stored before operation. During runtime, the system queries this pre-computed model for spatial feature locations rather than regenerating it, eliminating the need for continuous model reconstruction and reducing processing time.
3Measurement precision
If more spatial features are detected and used for pose estimation, then the localization accuracy improves, but the complexity of feature detection and matching increases
Solution Approach 1:
The environmental model is segmented into discrete spatial features with known locations. The system detects and matches only these predefined features rather than processing all visual data points, which simplifies the matching process while maintaining high accuracy through the use of multiple segmented features.
Data Source
AI summary
In one implementation, a method of localizing a device is performed at a device including one or more processors and non-transitory memory. The method includes obtaining an estimate of a pose of the device in an environment. The method includes obtaining an environmental model of the environment including a spatial feature in the environment defined by a first spatial feature location. The method includes determining a second spatial feature location of the spatial feature based on the estimate of the pose of the device. The method includes determining an updated estimate of the pose of the device based on the first spatial feature location and the second spatial feature location.