Autonomous Vehicle Sensor Calibration via Static Object Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The extrinsic parameter calibration of cameras and lidars in autonomous driving systems is complex and requires specialized equipment, making it unsuitable for automatic calibration during the driving process, leading to potential inaccuracies due to looseness issues.
Innovation Solution
A sensor calibration method and device that detects surrounding objects, recognizes static objects, and performs feature extraction using both cameras and lidars to calibrate extrinsic parameters in real-time during vehicle travel, simplifying the calibration process and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If extrinsic parameter calibration is performed using traditional methods with special calibration objects or in calibration rooms, then calibration accuracy can be maintained, but the operation becomes complex and requires frequent factory returns
Solution Approach 1:
The system performs self-calibration by utilizing naturally occurring static objects in the environment (buildings, trees, signs) as calibration targets. The camera and lidar automatically detect and extract features from these objects, and the system computes extrinsic parameters without requiring external calibration equipment or professional calibration facilities, enabling the vehicle to calibrate itself during normal operation
Solution Approach 2:
The calibration system uses universally available static objects in any environment rather than specialized calibration objects. The same camera and lidar sensors used for autonomous driving perception are also used for calibration purposes, eliminating the need for dedicated calibration equipment and allowing calibration to be performed anywhere the vehicle travels
2Reliability
If traditional calibration methods are used requiring factory returns, then calibration can be performed with proper equipment, but productivity and convenience are significantly reduced
Solution Approach 1:
The calibration process occurs continuously during normal vehicle operation rather than requiring periodic interruptions for factory returns. The system constantly detects static objects and performs feature extraction and parameter calibration in real-time, transforming calibration from a discrete maintenance task into an ongoing operational process that improves efficiency without compromising reliability
Solution Approach 2:
The patent replaces the mechanical process of physically transporting vehicles to calibration facilities with an automated computational system. The calibration function is substituted by algorithms that process sensor data and compute extrinsic parameters automatically during driving, eliminating the need for mechanical intervention and facility visits
3Adaptability or versatility
If camera and lidar are allowed to move or loosen during driving, then installation flexibility is improved, but measurement precision deteriorates due to positioning errors
Solution Approach 1:
The system continuously monitors the relative positions of camera and lidar by detecting static environmental objects and comparing their positions across multiple frames. When deviations are detected indicating loosening or movement, the system automatically computes correction parameters and adjusts the coordinate transformation relationships, providing real-time feedback that compensates for installation changes and maintains measurement precision
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
Embodiments of the present disclosure provide a sensor calibration method, a sensor calibration device, a computer device, a storage medium, and a vehicle. The method includes: detecting surrounding objects in a travelling process of a vehicle; recognizing a static object from the surrounding objects; performing feature extraction on the static object by a camera and a lidar, respectively; and calibrating an extrinsic parameter of the camera and the lidar based on the extracted feature.