Fiducial Objects: Custom Design and Evaluation

Area of Science:

• Computer Vision
• Robotics
• Geometric Measurement

Background:

• Camera pose estimation is crucial for robotics, medical imaging, and augmented reality.
• Existing fiducial markers like ArUco and Apriltag have limitations in accuracy and viewing angle when used individually.
• Custom fiducial objects improve visibility and precision by attaching markers to 3D structures.

Purpose of the Study:

• To develop novel custom-shaped fiducial markers for enhanced space utilization and detectability.
• To introduce a precise configuration estimation technique for these custom fiducial objects using multiviewpoint imaging.
• To provide open-source code, tutorials, and an application for creating and calibrating these objects.

Main Methods:

• Designing custom-shaped fiducial markers to optimize face coverage on 3D objects.
• Implementing a multiviewpoint imaging technique for precise object configuration estimation.
• Conducting empirical analysis to evaluate marker performance under various conditions (noise, blur, scale).

Main Results:

• Custom-shaped markers demonstrate superior space utilization compared to traditional square markers.
• The proposed method achieves precise configuration estimation of custom fiducial objects.
• Empirical results show significant performance improvements of custom markers over traditional ones in pose estimation tasks.

Conclusions:

• Custom-shaped fiducial markers represent a significant advancement in fiducial marker-based pose estimation.
• The developed techniques enhance marker detectability, space utilization, and overall pose estimation accuracy.
• The provided resources facilitate wider adoption and development in related research areas.