Vehicle LiDAR Building Outline Extraction Amid Bushes and Glass Reflections
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle lidar technologies face limitations in accurately extracting the outline of stationary objects, particularly in environments where buildings and bushes are mixed, and when lidar data is reflected through glass windows, leading to errors in object recognition and map matching.
Innovation Solution
A method for extracting building outlines using lidar data involves determining object points, establishing a reference point based on global and local quadrants, and identifying outer points through preprocessing and clustering, with adjustments for noise points and threshold distances to enhance accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a center of front bumper is used as reference point for object recognition, then the recognition process is simplified, but the outline extraction precision deteriorates when buildings and bushes are mixed
Solution Approach 1:
The patent divides the object recognition space into multiple quadrants (first, second, third, fourth quadrants) around the reference point. By segmenting the spatial domain, the system can process different regions independently and identify the convex hull for each quadrant separately, improving outline extraction precision while maintaining operational simplicity through systematic processing.
Solution Approach 2:
The patent introduces a quadrant-based spatial dimensioning system around the reference point, transforming the 2D plane into four distinct angular regions. This dimensional organization allows the system to differentiate between building outlines and bush outlines by analyzing point distribution patterns in different quadrants, thereby resolving the precision issue when objects are mixed.
2Productivity
If lidar data is processed using traditional methods, then processing speed is maintained, but reliability of building outline extraction deteriorates due to glass window reflections
Solution Approach 1:
The patent applies different processing criteria to different quadrants and regions. By determining convex hulls separately for each quadrant and analyzing local point distribution characteristics, the system can identify and filter out erroneous reflection points from glass windows while maintaining overall processing efficiency through localized rather than global reprocessing.
Solution Approach 2:
The patent creates a virtual convex hull model based on the distribution of object points in each quadrant. This virtual model serves as a reference to compare against actual lidar points, allowing the system to identify and eliminate reflection artifacts from glass windows by detecting points that deviate from the expected convex hull pattern, thereby improving reliability without sacrificing processing speed.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables the extraction of highly reliable building outlines, minimizing errors in object recognition and improving matching with precise map data by distinguishing between building and bush outlines and filtering out internal reflections.
Implementation Method 1
light detection and ranging (lidar) sensor
Data Source
AI summary
An object outline extracting method for a vehicle includes: as light detection and ranging (lidar) data is received, determining a reference point based on positions of points in global quadrants generated based on the vehicle and local quadrants generated based on a cluster box of an object, and detecting the object; extracting representative points of the points and outer points of the object; determining a position of a second point with respect to a first straight line connecting a first point and a third point among the outer points; and determining the second point as an outer point of the object or as a noise point based on the position of the second point with respect to the first straight line.