Multi-Camera Calibration Using Dense Depth Maps Without Markers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional sensor calibration techniques for autonomous vehicles require infrastructure like fiducial markers, leading to downtime and potential safety hazards due to misalignment, which can occur during normal driving conditions and are often time-consuming, taking hours to complete.
Innovation Solution
The development of techniques to calibrate sensors without infrastructure, using methods such as generating dense depth maps from sparse point cloud data and projecting meshes into a two-dimensional camera space to identify and quantify misalignment between sensors, allowing for rapid calibration within minutes or seconds, enabling real-time data refinement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional calibration techniques using fiducial markers are used, then calibration accuracy can be achieved, but system downtime increases and safety risks arise due to the need to travel to calibration locations
Solution Approach 1:
The patent extracts the calibration function from infrastructure-dependent methods to infrastructure-independent methods by using natural environment features (buildings, roads, sidewalks) as calibration references instead of fiducial markers, allowing calibration to occur anywhere without requiring specialized calibration locations
Solution Approach 2:
The system performs self-calibration by using its own sensor data and processing capabilities to automatically determine calibration parameters through optimization algorithms, eliminating the need for external calibration infrastructure and manual intervention
2Ease of manufacture
If infrastructure-based calibration is used, then calibration can be performed at manufacturing locations, but subsequent calibration requires traveling to locations with infrastructure, creating safety hazards
Solution Approach 1:
The patent removes the dependency on specialized calibration infrastructure by extracting the essential calibration function and implementing it through natural environment features that are universally available, eliminating the need to travel to specific calibration locations
Solution Approach 2:
The system changes the calibration approach from using fixed infrastructure parameters (fiducial markers at specific locations) to using variable environmental parameters (natural features in any location), allowing calibration to adapt to any environment without safety risks
3Reliability
If traditional calibration methods are used, then calibration can be performed, but it takes hours to complete, reducing productivity
Solution Approach 1:
The patent implements periodic calibration checks during normal sensor operation rather than requiring dedicated calibration time, allowing the system to continuously refine calibration parameters through repeated optimization cycles using incoming sensor data
Solution Approach 2:
The calibration process continues continuously during normal operation rather than stopping the system, with the optimization algorithm constantly refining calibration parameters using real-time sensor data from the environment
Data Source
AI summary
This disclosure is directed to calibrating sensor arrays, including sensors arrays mounted on an autonomous vehicle. Image data from multiple cameras in the sensor array can be projected into other camera spaces using one or more dense depth maps. The dense depth map(s) can be generated from point cloud data generated by one of the sensors in the array. Differences determined by the comparison can indicate alignment errors between the cameras. Calibration data associated with the errors can be determined and used to calibrate the sensor array without the need for calibration infrastructure.