Ambiguity-Free Optical-Inertial Tracking for Augmented Reality Headsets

Area of Science:

• Computer Vision
• Robotics
• Human-Computer Interaction

Background:

• Advancements in computing power enable self-contained augmented reality (AR) headsets with head-anchored tracking.
• Infrared marker tracking offers accuracy but wearable visible-light cameras struggle with tracking failures under uncontrolled lighting.

Purpose of the Study:

• To improve the accuracy and robustness of optical self-tracking in AR systems.
• To address limitations caused by marker occlusions, poor camera calibration, and inconsistent lighting.

Main Methods:

• A sensor fusion approach using Kalman filtering to integrate optical and inertial tracking data.
• Development of a custom AR headset for complex manual tasks under direct vision.
• Experimental evaluation of AR overlay accuracy.

Main Results:

• The proposed sensor fusion method improves head-mounted display (HMD) tracking accuracy by one third.
• Enhanced robustness in capturing target scene orientation during marker occlusions.
• Mitigation of tracking instabilities caused by uncontrollable lighting conditions.

Conclusions:

• Sensor fusion effectively overcomes the limitations of purely optical tracking in AR systems.
• The approach provides more reliable pose estimation for AR applications in real-world environments.
• Improved tracking performance supports complex manual tasks requiring high AR overlay precision.