LiDAR Layer Fusion for Reliable Object Shape and Heading Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The reliability of autonomous driving systems equipped with LiDAR sensors is compromised when incorrect information about target vehicles is acquired, particularly due to inappropriate selection of layers in multi-layered point data, leading to deteriorated performance and accuracy in determining the shape and heading of objects.
Innovation Solution
A method and device using a LiDAR sensor to analyze the shape and heading of objects by determining shape flags for individual layers and a whole layer, with reliability conditions to adjust or maintain the shape flag, including priority rules, score thresholds, and distance differences, ensuring accurate object tracking.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple-layered point data are obtained through LiDAR sensor and heading is determined from a selected layer, then the shape and heading information can be acquired, but if the layer selection is not appropriate, the reliability deteriorates greatly
Solution Approach 1:
The patent divides the LiDAR point cloud data into multiple layers (first layer, second layer, third layer, etc.) based on vertical height positions. Each layer is independently analyzed to determine shape flags (L-shape, I-shape, sL-shape, or unknown), and these individual layer results are then integrated to determine the final object shape. This segmentation allows the system to process complex 3D data into manageable 2D projections, improving both measurement precision and reliability by avoiding inappropriate single-layer selection.
Solution Approach 2:
The patent merges the shape determination results from multiple individual layers to arrive at a final conclusive shape flag for the object. By combining information from multiple layers rather than relying on a single selected layer, the system achieves more reliable shape and heading determination, directly resolving the technical contradiction between measurement precision and system reliability.
2Loss of information
If LiDAR points of each layer are analyzed individually to determine shape flags, then detailed shape information can be obtained, but the complexity of data processing increases
Solution Approach 1:
The patent segments the complex 3D LiDAR data into multiple 2D layers, where each layer can be independently processed to determine its shape flag. This segmentation simplifies the processing of individual layers while maintaining complete shape information through the integration of all layer results, effectively managing data processing complexity without losing shape information.
Solution Approach 2:
The patent transforms 3D LiDAR point cloud data into 2D layer projections by dividing the space into multiple horizontal layers. This dimensionality change from 3D to 2D simplifies the shape analysis process for each layer while preserving the complete 3D shape information through the combination of all 2D layer results, reducing processing complexity without information loss.
Data Source
AI summary
The present disclosure relates to a method of analyzing a shape of an object by use of LiDAR and a device for tracking an object according to the same. A method for analyzing a shape of an object by use of LiDAR, according to an embodiment of the present disclosure, comprises obtaining first to Mth (M is an integer of 2 or greater) layers of LiDAR points spaced apart in a vertical direction with respect to an object by the LiDAR sensor, obtaining LiDAR points of a whole layer by projecting whole LiDAR points for the object or LiDAR points of the first to Mth layers onto the whole layer, and determining shape flags for the first to Mth layers and the whole layer, respectively, each by use of at least a part of corresponding LiDAR points of each layer according to a plurality of predetermined shape types, and determining a shape flag of the object.