LiDAR Object Recognition via Point Cloud Layer Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles and vehicles with driver assistance systems face challenges in accurately determining whether an object represented by a point cloud is occluding or being occluded by another object, which affects their navigation and obstacle avoidance capabilities.
Innovation Solution
An object recognition apparatus and method using LiDAR, which identifies key points in a point cloud and compares reliability values based on distances and angles to determine occlusion relationships between objects, employing a processor to assign reliability values and identify occluding or occluded objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Difficulty of detecting and measuring
If LiDAR is used to obtain point cloud data for object detection, then the ability to detect surrounding environments and distinguish obstacles is improved, but the accuracy in determining occlusion relationships between objects deteriorates
Solution Approach 1:
The patent segments the point cloud data into multiple layers based on distance from the host vehicle, creating front, middle, and rear layers. This segmentation allows the system to systematically analyze spatial relationships and determine occlusion status by comparing points across different layers, thereby improving occlusion determination accuracy while maintaining comprehensive object detection capability
Solution Approach 2:
The patent introduces a layered dimensional structure to the point cloud analysis, organizing points by their distance from the host vehicle into multiple layers. This dimensional transformation enables the system to resolve occlusion relationships by analyzing the spatial distribution of points across layers, converting a complex 3D occlusion problem into a more manageable multi-layer comparison problem
2Reliability
If the system identifies occlusion relationships to improve navigation accuracy, then the navigation and obstacle avoidance capabilities are improved, but the computational complexity and processing time increase
Solution Approach 1:
The patent divides the point cloud into distance-based layers and processes each layer separately, identifying occlusion relationships within and between layers. This segmented approach reduces computational complexity by breaking down the overall problem into smaller, manageable sub-problems while maintaining reliable navigation accuracy through systematic analysis
Solution Approach 2:
The patent focuses on identifying key external points (frontmost and rearmost points) in each layer rather than processing all points comprehensively. This partial action approach maintains sufficient accuracy for navigation decisions while significantly reducing computational burden by concentrating analysis on critical points that determine occlusion relationships
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 accuracy of determining occlusion relationships between objects, enhancing the navigation and obstacle avoidance capabilities of autonomous vehicles and driver assistance systems.
Implementation Method 1
A distance from a LiDAR to an object may be obtained through an interval between the time when laser is transmitted by the LiDAR and the time when the laser reflected by the object is received
Data Source
AI summary
In an object recognition apparatus and method, the object recognition apparatus includes a LiDAR and a processor. The processor may identify a first point, a second point, a third point, and a fourth point, identify a first external point, and a second external point, identify an angle between a line segment connecting the first external point and the LiDAR and a line segment connecting the second external point and the LiDAR identify, as an occluded object, an object which is one of the first object and the second object, when the angle is less than or equal to a threshold angle, and identify an object different from the object identified as the occluded object among the first object and the second object as an object occluding the occluded object.