UAV Radar-Camera Calibration for Height and Pitch Changes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data fusion methods for unmanned aerial vehicles fail to accurately integrate millimeter-wave radar and visual data due to the inability to adapt to changes in height and pitch angle, resulting in significant deviations in depth information.
Innovation Solution
A joint calibration method that establishes a three-dimensional data entry model incorporating ground height and pitch angle changes, using calibration parameters to determine a spatial conversion relationship between the millimeter-wave radar and image acquisition device, enabling accurate data fusion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional data fusion methods are used without considering pose changes, then the system complexity is low, but the measurement precision of depth information deteriorates significantly
Solution Approach 1:
The patent pre-establishes a calibration parameter set containing multiple groups of calibration parameters corresponding to different pose information intervals (height and pitch angle ranges) before actual operation. When the detection radar's pose changes during operation, the system simply selects the pre-computed calibration parameters matching the current pose interval, avoiding complex real-time calculations and maintaining both high precision and low complexity.
Solution Approach 2:
The patent introduces dynamic adaptation by dividing the operating range into multiple pose information intervals and selecting different calibration parameter groups based on the current height and pitch angle. This dynamic selection mechanism allows the system to adapt to pose changes without requiring complex real-time recalculation, resolving the contradiction between precision and complexity.
2Adaptability or versatility
If calibration parameters are fixed without adapting to pose changes, then the device complexity is low, but the adaptability to different flight conditions deteriorates
Solution Approach 1:
The patent segments the continuous operating range into discrete pose information intervals based on height and pitch angle ranges. Each interval has its own pre-computed calibration parameters. This segmentation allows the system to adapt to different flight conditions by selecting the appropriate segment, achieving high adaptability without requiring complex continuous adjustment mechanisms.
Solution Approach 2:
The system performs calibration computations in advance for all possible pose intervals and stores the results in a calibration parameter set. During operation, it only needs to select the pre-computed parameters matching the current pose interval, achieving adaptability to various flight conditions while keeping the operational complexity low.
3Reliability
If real-time calibration recalculation is performed when pose changes, then the adaptability is high, but the computing time and processing speed increase significantly
Solution Approach 1:
The patent performs all calibration computations in advance and stores multiple groups of calibration parameters corresponding to different pose intervals. When the detection radar's pose changes during operation, the system simply selects the pre-computed parameters matching the current pose interval, avoiding time-consuming real-time calculations and ensuring both high reliability and fast response.
Solution Approach 2:
The system dynamically selects from pre-computed calibration parameters based on current pose conditions, achieving adaptive calibration without real-time computation. This dynamic selection from a pre-prepared set maintains data fusion accuracy while minimizing computation time.
Data Source
AI summary
A joint calibration method is implemented in an unmanned aerial vehicle. In the method, pose information of a detection radar is obtained, and the pose information includes a ground height of the detection radar and a pitch angle of the detection radar. Target calibration parameters matching the pose information in a preset calibration parameter set are obtained. The calibration parameter set includes groups of calibration parameters, and each group of calibration parameters matching a pose information interval and the pose information interval includes a height interval and a pitch angle interval. A spatial conversion relationship between the detection radar and an image acquisition device is determined based on the target calibration parameters.