Extrinsic Sensor Calibration Using Nonparallel Target Corners
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current sensor fusion methods for vehicles require extensive computational resources for extrinsic calibration, often introducing reprojection errors and inefficiencies due to data association processes.
Innovation Solution
A computationally efficient system and process for extrinsic sensor calibration using a closed-form solution that estimates relative positions and rigid transformations between sensors, eliminating the need for data association by employing targets with nonparallel flat surfaces and reflectors, and utilizing least squares optimization for accurate alignment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If data association processes are used for extrinsic calibration, then sensor alignment can be achieved, but computational resources are excessively consumed and computational steps are increased
Solution Approach 1:
The patent extracts and eliminates the data association process from the extrinsic calibration pipeline. By using a closed-form solution that directly computes the transformation matrix from corresponding points between two sensors, the method removes the computationally intensive iterative matching step while maintaining calibration accuracy.
Solution Approach 2:
The patent replaces the iterative optimization mechanism (data association with reprojection error minimization) with a direct closed-form mathematical solution. This substitution uses linear algebra operations to compute the essential matrix and fundamental matrix directly, avoiding the need for iterative numerical optimization.
2Measurement precision
If data association processes are used for extrinsic calibration, then sensor alignment can be achieved, but the process becomes inefficient and time-consuming
Solution Approach 1:
The patent performs preliminary actions by establishing point correspondences between two sensors through direct geometric relationships rather than iterative matching. The method pre-computes the essential matrix from matched points and then directly solves for the transformation matrix, eliminating the need for time-consuming iterative optimization during the calibration process.
Solution Approach 2:
The patent skips the iterative optimization steps typically required in data association-based calibration. By using a closed-form solution, the method rushes through the calibration process in a single computational pass, directly computing the transformation matrix without repeatedly refining the solution through iterative reprojection error minimization.
3Adaptability or versatility
If conventional sensor fusion methods are used, then data from multiple sensors can be combined, but reprojection errors are introduced reducing accuracy
Solution Approach 1:
The patent converts the potential harm of reprojection errors into a benefit by using a different mathematical approach that avoids them entirely. Instead of minimizing reprojection errors through iterative optimization, the method uses closed-form solutions based on direct geometric relationships, transforming the calibration problem into one that is inherently free from reprojection error accumulation.
Data Source
AI summary
An extrinsic-calibration system includes at least one target and a computer. Each target includes three flat surfaces that are mutually nonparallel and a corner at which the three surfaces intersect. The computer is programmed to estimate a first set of relative positions of the corner in a first coordinate frame from a first sensor, the first set of relative positions corresponding one-to-one to a set of absolute positions of the at least one target; estimate a second set of relative positions of the corner in a second coordinate frame from a second sensor, the second set of relative positions corresponding one-to-one to the set of absolute positions; and estimate a rigid transformation between the first coordinate frame and the second coordinate frame based on the first set of relative positions and the second set of relative positions.