Multi-Sensor Extrinsic Calibration Using Filtered Point Regions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern vehicles with multiple sensors face challenges in accurately calibrating sensors of different modalities, leading to potential erroneous calibrations and reduced reliability, especially when extrinsic calibration is required to match points from one sensor to another.
Innovation Solution
A method that involves obtaining data from multiple sensors, filtering irrelevant points based on specific regions and properties, and determining calibration parameters using a priori knowledge and detected objects to improve accuracy and reliability, allowing for the calibration of multiple sensors simultaneously.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all data points from multiple sensors are used for calibration, then the calibration process can be performed with available data, but the reliability of calibration deteriorates due to inclusion of irrelevant points
Solution Approach 1:
The patent extracts and removes irrelevant data points from the sensor data before performing calibration. The system identifies and filters out points that do not contribute to accurate calibration, keeping only the relevant points for the calibration process. This extraction principle directly resolves the contradiction by reducing the quantity of data points to only those that are useful, thereby improving calibration reliability.
Solution Approach 2:
The patent applies different quality standards to different regions of sensor data. By defining specific regions of interest and applying localized filtering criteria, the system ensures that only data points within these regions are used for calibration. This local quality approach improves reliability by ensuring that only high-quality, relevant data points from specific spatial regions are included in the calibration process.
2Reliability
If filtering is applied to remove irrelevant points, then the reliability of calibration is improved, but the complexity of the calibration process increases
Solution Approach 1:
The patent performs filtering of irrelevant data points as a preliminary action before the actual calibration process. By pre-processing the sensor data to remove irrelevant points beforehand, the system simplifies the subsequent calibration steps. This preliminary filtering action reduces the complexity of the overall calibration process while maintaining high reliability, as the calibration algorithm works with already-prepared, high-quality data.
3Productivity
If a priori calibration is used to convert data into a common domain, then the calibration process is accelerated, but the accuracy may deteriorate due to initial approximation errors
Solution Approach 1:
The patent uses a priori calibration as a preliminary action to quickly convert sensor data into a common domain before performing the final accurate calibration. The system first applies the approximate a priori transformation to establish a common reference frame, then refines the calibration parameters using only relevant filtered data points. This two-stage approach maintains high productivity by leveraging the speed of a priori calibration while improving accuracy through subsequent refinement with filtered data.
Solution Approach 2:
The patent implements a feedback mechanism where the initial a priori calibration results are used to guide the subsequent filtering and refinement process. The system evaluates the preliminary calibration outcomes and uses this feedback to adjust the filtering criteria and refinement steps, ensuring that the final calibration achieves high accuracy while maintaining the productivity benefits of the initial fast transformation.
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
The present invention provides a method for calibrating a first and a second sensor of a vehicle, the method comprising: - obtaining first data from the first sensor and second data from the second sensor, - filtering at least the second data based on positions of data points of the second data, and - determining one or more parameters of a calibration between the first and the second sensor based on the first data and the filtered second data.