Gaze Axis Correction for XR Eye Tracking Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing gaze tracking systems in extended reality (XR) devices suffer from errors due to the offset between the actual visual axis and the modeled optical axis of the human eye, leading to inaccuracies in displaying virtual content and determining the user's gaze direction.
Innovation Solution
A method is implemented to correct the gaze direction and origin by determining the true visual axis during enrollment, accounting for clip-on lenses if present, and applying point of view corrections to align the gaze estimation with the display, using eye tracking technology and optical characteristics to adjust virtual content display accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If the optical axis is used in the gaze estimation process to determine the point on the display at which the user is looking, then the gaze tracking system can operate with a simple eye modeling method, but the accuracy of gaze determination is reduced due to the offset between the actual visual axis and the modeled optical axis
Solution Approach 1:
The patent applies preliminary action by performing eye modeling calibration during the enrollment phase before actual gaze tracking. The system pre-determines the relationship between the optical axis and visual axis through captured images and stored models, so that during operation, the pre-calibrated model can be used to accurately estimate gaze without real-time complex calculations. This resolves the contradiction by preparing the correction data in advance, maintaining simple real-time operation while achieving accurate gaze determination.
Solution Approach 2:
The patent changes parameters by transforming the eye model from using only the optical axis to incorporating both the optical axis and the visual axis parameters. The system captures images during enrollment to determine the actual visual axis position relative to the optical axis, storing these parameters for later use. This parameter transformation allows the system to account for the offset between axes, improving gaze accuracy without significantly increasing operational complexity.
2Ease of manufacture
If the entrance pupil determined by the eye modeling method is used for rendering virtual content, then the rendering process is simplified, but errors occur in displaying virtual content due to the difference between the actual entrance pupil and the modeled entrance pupil
Solution Approach 1:
The patent applies preliminary action by determining the correct entrance pupil position during the enrollment phase through captured images and storing it in the eye model. During virtual content rendering, the system uses this pre-determined entrance pupil position rather than calculating it in real-time. This approach maintains rendering process simplicity while ensuring display accuracy, as the correct entrance pupil parameters are prepared in advance and applied during content generation.
3Measurement precision
If clip-on lenses are added to correct vision, then the user's visual acuity is improved, but the complexity of the optical system increases and requires additional correction calculations for gaze tracking
Solution Approach 1:
The patent applies preliminary action by capturing images during enrollment with the clip-on lenses already in place. This allows the system to determine the entrance pupil and optical characteristics specific to the user's corrected vision state before actual use. By pre-calibrating the eye model with the lenses present, the system accounts for the additional optical complexity without requiring real-time calculations, thus maintaining operational simplicity while achieving accurate gaze tracking with corrected vision.
Solution Approach 2:
The patent uses the enrollment process as an intermediary that bridges the gap between the added optical complexity of clip-on lenses and the simplicity required for operational gaze tracking. The enrollment phase serves as a mediator where the system captures and stores the specific optical characteristics of the user's eye system with lenses, transforming the complex physical setup into a simplified computational model for subsequent use.
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 enhances the accuracy of gaze tracking in XR devices by aligning the visual axis with the display origin, reducing errors in virtual content rendering and ensuring precise gaze determination.
Implementation Method 1
one or more infrared (IR) light sources emit IR light towards a user's eye. A portion of the IR light is reflected off the eye and captured by an eye tracking camera
Data Source
AI summary
Methods and apparatus for correcting the gaze direction and the origin (entrance pupil) in gaze tracking systems. During enrollment after an eye model is obtained, the pose of the eye when looking at a target prompt is determined. This information is used to estimate the true visual axis of the eye. The visual axis may then be used to correct the point of view (PoV) with respect to the display during use. If a clip-on lens is present, a corrected gaze axis may be calculated based on the known optical characteristics and pose of the clip-on lens. A clip-on corrected entrance pupil may then be estimated by firing two or more virtual rays through the clip-on lens to determine the intersection between the rays and the corrected gaze axis.