LiDAR Object Detection Using Height-Based Cluster Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
LiDAR systems face inaccuracies in object detection due to crosstalk, which can jeopardize autonomous driving safety by generating point data in regions where no object actually exists, especially when sensing objects with high reflectivity like road signs.
Innovation Solution
The method involves generating overhead and grounded clusters in point cloud data based on object height, using mesh graphs and grid maps to differentiate and remove erroneous clusters caused by crosstalk, thereby improving recognition accuracy for objects with high reflectivity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If LiDAR sensor senses objects with high reflectivity, then detection coverage is improved, but crosstalk generates false point data in regions where no object exists
Solution Approach 1:
The patent segments point cloud data into overhead clusters (objects above reference height) and grounded clusters (objects at or below reference height) based on height information. This segmentation allows independent processing and comparison of the two cluster types to identify and remove crosstalk-generated false points while preserving valid detection data.
Solution Approach 2:
The patent introduces height information from a reference object as an intermediary parameter to distinguish between valid overhead objects and false grounded objects caused by crosstalk. By using height as a discriminating factor, the system can identify and remove erroneous point data without affecting the detection of actual objects.
2Measurement precision
If point cloud data is processed to remove crosstalk, then recognition accuracy is improved, but processing complexity increases
Solution Approach 1:
The patent divides the point cloud data processing into distinct segments: overhead cluster generation, grounded cluster generation, and comparison/ removal operations. This segmentation simplifies the overall processing complexity by breaking down the complex task into manageable, modular steps that can be executed sequentially.
Solution Approach 2:
The patent changes the processing approach by introducing height-based parameter thresholds (reference height) to automatically differentiate between overhead and grounded clusters. This parameter-based approach simplifies the removal of crosstalk data compared to more complex methods that would require additional sensors or sophisticated algorithms.
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 enhances the accuracy of recognizing overhead objects like road signs by removing clusters generated due to crosstalk, maximizing object recognition performance within limited system resources and ensuring safer autonomous driving.
Implementation Method 1
LiDAR (Light Detecting And Ranging) has been developed in the form of constructing topographic data
Implementation Method 2
sensing an object with high reflectivity
Data Source
AI summary
An object detecting method of a vehicle LiDAR system may be disclosed. The object detecting method includes generating an overhead cluster corresponding to an object whose height from the ground may be equal to or larger than a reference height and a grounded cluster whose height from the ground may be smaller than the reference height, in point cloud data obtained by sensing an object; and comparing point data included in the overhead cluster and point data included in the grounded cluster, and removing the grounded cluster upon determining that an object corresponding to the corresponding grounded cluster does not exist.