Vehicle LiDAR Planar Point Detection for Non-Beam Point Clouds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing feature extraction techniques for lidar point clouds are unsuitable for non-beam-based point clouds, limiting the identification of planar points and affecting the accuracy of object detection in semi-autonomous or autonomous vehicle operations.
Innovation Solution
A method involving principal component analysis (PCA) to calculate normal vectors, followed by RANSAC and DBSCAN algorithms, is used to identify horizontal and vertical feature points in both beam-based and non-beam-based lidar systems, enhancing the detection of planar points in lidar point clouds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing feature extraction techniques are used, then processing speed is maintained, but identification accuracy of planar points deteriorates for non-beam-based point clouds
Solution Approach 1:
The patent changes the processing parameters by calculating normal vectors for each point using PCA and then applying threshold-based filtering on the normal vector components. This parameter transformation enables the system to identify planar points in non-beam-based point clouds by converting spatial relationship information into directional information that can be universally processed regardless of point cloud generation method
Solution Approach 2:
The patent creates a universal feature extraction method that works for both beam-based and non-beam-based point clouds. By using normal vector calculation and component thresholding rather than beam-specific processing, the system achieves multi-functionality across different lidar types while maintaining high identification accuracy for planar points
2Measurement precision
If normal vector calculation with PCA is performed for each point, then identification accuracy improves, but computational complexity increases
Solution Approach 1:
The patent extracts only the essential information needed for planar point identification by calculating normal vectors through PCA and then extracting points that satisfy specific threshold conditions on normal vector components. This extraction approach focuses computational resources on the most relevant features rather than processing all point cloud data equally, reducing overall complexity while maintaining accuracy
3Measurement precision
If threshold filtering on normal vector components is applied, then horizontal and vertical feature points are accurately identified, but processing time increases
Solution Approach 1:
The patent applies partial action by using threshold filtering on only the dominant components of normal vectors (e.g., |nx| < threshold and |nz| > threshold for horizontal surfaces) rather than analyzing all three components equally. This partial processing approach identifies the majority of planar points efficiently while avoiding unnecessary computation on points that clearly don't meet the criteria, reducing processing time while maintaining identification accuracy
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
Improves the identification of planar points in lidar point clouds, thereby improving the accuracy of object detection and facilitating better semi-autonomous or autonomous vehicle operations.
Implementation Method 1
a lidar system to transmit incident light and receive reflections from one or more objects as a point cloud of points
Data Source
AI summary
A system in a vehicle includes a lidar system to transmit incident light and receive reflections from one or more objects as a point cloud of points. The system also includes processing circuitry to identify feature points among the points of the point cloud, the feature points being horizontal feature points reflected from a horizontal surface or vertical feature points reflected from a vertical surface. The processing circuitry processes the point cloud by obtaining a normal vector corresponding to each of the points of the point cloud. The normal vector includes a first component associated with a first dimension, a second component associated with a second dimension, and a third component associated with a third dimension.