LiDAR Object Classification for Separating Merged Object Contours
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object classification systems for autonomous vehicles struggle to accurately separate and identify multiple objects that are recognized as a merged object, leading to poor tracking performance and increased risk of system performance degradation.
Innovation Solution
The proposed object classification apparatus and method utilize a LiDAR and a processor to identify contour points that satisfy specific distribution, dispersion, and shape conditions, allowing for the separation and clustering of contour points into distinct objects, and efficiently manage memory spaces for storing object information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If multiple objects are recognized as a merged object, then the object detection system can maintain simplicity in processing, but the tracking performance and object identification accuracy deteriorate
Solution Approach 1:
The patent applies segmentation by dividing the merged object's contour points into multiple distinct object groups based on spatial distribution characteristics. The processor analyzes the geometric arrangement of contour points and separates them into individual object regions, enabling accurate identification of multiple objects that were initially detected as a single merged object.
2Measurement precision
If contour points are separated into multiple objects, then object identification accuracy improves, but the system complexity and computational load increase
Solution Approach 1:
The patent changes parameters by evaluating multiple geometric characteristics of contour points including distribution patterns, dispersion metrics, and shape features. The processor varies these parameter thresholds dynamically to determine the optimal separation criteria, balancing accurate object separation with computational efficiency without requiring overly complex processing 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 improves the accuracy of object separation, enhances tracking performance for multiple objects, and reduces the risk of system performance degradation by effectively identifying and managing object information in memory spaces.
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
An object classification apparatus includes a LiDAR and a processor. The processor may identify that a point corresponding to a first previous object box and a point corresponding to a second previous object box are included in an integrated object box including contour points at the predetermined time and included in an object box representing an integrated object, separate and cluster the contour points at the predetermined time into contour points representing a first object and contour points representing a second object, store the separated contour points representing the first object in association with the first object, and store the separated contour points representing the second object in association with the second object based on a number of separated contour points representing the first object and a number of separated contour points representing the second object.