Periodic Parameter Calibration for Visual-Inertial Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing visual-inertial tracking systems in AR/VR devices suffer from parameter drift due to mechanical stress and temperature changes, leading to prolonged convergence times and reduced tracking accuracy, which affects power consumption and computational resources.
Innovation Solution
Implement a method where the tracking system is periodically operated to determine parameter values, storing the latest values for immediate reuse, reducing convergence time and improving accuracy by starting from an up-to-date estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the tracking system operates continuously to maintain accurate parameter values, then tracking accuracy is improved, but power consumption and computational resource usage increase
Solution Approach 1:
The system implements periodic calibration operations at predetermined time intervals rather than continuous operation. The processor periodically determines updated calibration parameters by processing sensor data, maintaining tracking accuracy while allowing periods of lower computational activity that reduce power consumption and resource usage between calibration events.
Solution Approach 2:
The system performs calibration parameter determination in advance at scheduled intervals, storing updated parameters for future use. This preliminary action ensures accurate parameters are ready before they are needed, allowing the system to operate efficiently with pre-computed values rather than continuously recalculating, thus reducing ongoing power consumption while maintaining accuracy.
2Loss of time
If the tracking system operates continuously to maintain up-to-date parameter values, then convergence time is reduced, but computational resource usage increases
Solution Approach 1:
The system performs computational resource-intensive calibration operations periodically at predetermined intervals rather than continuously. This approach maintains up-to-date parameter values ready for quick convergence when needed, while significantly reducing computational resource usage during intervals between periodic calibration events.
Solution Approach 2:
The system performs calibration parameter determination in advance and stores the results. When tracking begins or resumes, the system can quickly load pre-computed parameters, achieving fast convergence without requiring continuous computational resources. The preliminary calibration work is done during scheduled maintenance periods rather than during active tracking.
3Device complexity
If factory calibration parameters are used without updates, then device complexity is reduced, but parameter drift occurs leading to reduced tracking accuracy
Solution Approach 1:
The system maintains simplicity by using fixed factory calibration parameters during normal operation, reducing complexity. However, it implements periodic calibration operations at predetermined intervals to detect and correct parameter drift, thereby maintaining tracking accuracy without requiring continuous complex adaptive algorithms.
Solution Approach 2:
The system performs self-calibration by automatically processing sensor data and determining updated calibration parameters without user intervention. This self-service approach corrects parameter drift through periodic automated calibration, maintaining accuracy while avoiding the complexity of manual calibration procedures or continuous adaptive algorithms.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for calibrating a visual-inertial tracking system is described. A device operates the visual-inertial tracking system without receiving a tracking request from a virtual object display application. In response to operating the visual-inertial tracking system, the device accesses sensor data from sensors at the device. The device identifies, based on the sensor data, a first calibration parameter value of the visual-inertial tracking system and stores the first calibration parameter value. The system detects a tracking request from the virtual object display application. In response to the tracking request, the system accesses the first calibration parameter value and determines a second calibration parameter value from the first calibration parameter value.