Optical Head Pose Mapping With Multi-Density Fiducials
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing head tracking systems for head-worn displays face challenges in achieving high-integrity head pose estimation due to the reduced detectability of high-density fiducials, which are difficult to capture with sufficient pixel accuracy, leading to potential inaccuracies in safety-critical applications.
Innovation Solution
A system utilizing a combination of medium-density and low-density fiducials, where medium-density fiducials include encoded identifiers and low-density fiducials provide positional information, enabling high-integrity head pose estimation through a processor that determines the head pose by combining 2D image data with a 3D constellation database.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-density fiducials are used to increase correspondence integrity, then measurement precision is improved, but detectability deteriorates
Solution Approach 1:
The patent divides the fiducial system into multiple separate fiducials with medium-density encoding rather than using a single high-density fiducial. This segmentation allows each individual fiducial to remain detectable while the collection provides sufficient correspondence integrity through their combined unique identifiers and spatial relationships.
Solution Approach 2:
Different fiducials are assigned different density levels (medium-density with encoded identifiers versus low-density without encoded identifiers) based on their specific roles in the system. Medium-density fiducials provide identification while low-density fiducials provide positional reference, optimizing detectability for each type's function.
2Loss of information
If the density of bits within an individual fiducial is increased, then correspondence integrity is improved, but detectability deteriorates
Solution Approach 1:
The information encoding is segmented across multiple fiducials rather than concentrated in a single high-density fiducial. Each medium-density fiducial contains some encoded information, and the complete correspondence integrity is achieved through the combination of identifiers from multiple fiducials plus their spatial arrangement.
Solution Approach 2:
The system transitions from relying solely on high bit-density within individual fiducials to using spatial dimension information (positions and orientations of multiple fiducials in 3D space) as an additional layer of encoding. This dimensional approach maintains detectability while achieving correspondence integrity.
3Difficulty of detecting and measuring
If multiple fiducials are used instead of a single high-density fiducial, then detectability is improved, but device complexity increases
Solution Approach 1:
Multiple fiducials share common detection and processing mechanisms, allowing the system to handle multiple fiducials using the same algorithmic approach. This universality reduces the complexity increase that would otherwise result from having multiple distinct fiducial types.
Solution Approach 2:
The patent uses fiducials with similar visual characteristics and detection properties (medium-density encoding scheme) rather than mixing vastly different fiducial types. This homogeneity simplifies the detection and processing system while still achieving the needed correspondence integrity through multiple fiducials.
Data Source
Figure 1
Figure 2
Figure 3A~3B
AI summary
A system and method for high integrity head pose estimation receives a 2D image including medium-density fiducials (302) and low-density fiducials (304). The system then detects the medium-density fiducials (302) and compares them with a 3D constellation to form an initial head pose estimate. The initial head pose estimate is then checked for possible alternative correspondences. A selection from a set of candidate initial head pose estimates is made. The selected head pose estimate then has search areas established for the medium-density fiducials (302) and low-density fiducials (304). The system then determines whether or not the head pose is valid and accurate based on whether or not the fiducials appear in their respective search areas.