3D Lane Generation Using Depth-Aware Neural Networks

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing lane recognition technologies struggle to accurately identify and generate three-dimensional lane information, especially in diverse and complex road environments, due to limitations in understanding topographical features.

Innovation Solution

A neural network-based method that generates lane information by creating a lane probability map, extracting lane and depth feature information, generating depth distribution and spatial information, and calculating offset information to produce three-dimensional lane data.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If deep learning-based lane recognition is applied to diverse road environments, then lane recognition accuracy is improved, but understanding of topographical features is insufficient

Engineering Contradiction:

Improvelane recognition accuracyVSAvoidtopographical feature understanding

Core Design Contradiction:

Measurement precisionVSLoss of information

Solution Approach 1:

The patent transforms 2D lane detection into 3D lane information by introducing depth dimension. Multiple neural networks process the input image to generate lane probability maps, depth information, and spatial coordinates, converting planar lane recognition into three-dimensional spatial understanding that captures topographical features effectively

Inventive Principle:

Principle #17Another dimension (Dimensionality change)

Solution Approach 2:

The patent divides the lane recognition task into multiple independent processing streams using separate neural networks. One network generates lane probability maps, another extracts depth information, and a third calculates spatial coordinates. This segmentation allows each network to specialize in specific features while collectively achieving comprehensive topographical understanding

Inventive Principle:

Principle #1Segmentation

2Measurement precision

If three-dimensional lane information is generated using multiple neural networks, then lane recognition accuracy is improved, but device complexity increases

Engineering Contradiction:

Improvelane recognition accuracyVSAvoidneural network architecture complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent designs neural networks that perform multiple functions simultaneously. The same network architecture is used for generating both lane probability maps and depth information from the input image, reducing the need for entirely separate specialized networks and simplifying the overall system while maintaining high accuracy

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS12272158B2Method and apparatus with lane generation

Publication Date: 2025.04.08 SAMSUNG ELECTRONICS CO LTD

AI summary

A method of generating lane information using a neural network includes generating a lane probability map based on an input image, generating lane feature information and depth feature information by applying the lane probability map to a second neural network, generating depth distribution information by applying the depth feature information to a third neural network, generating spatial information based on the lane feature information and the depth distribution information, generating offset information including a displacement between a position of a lane and a reference line by applying the spatial information to a fourth neural network, and generating three-dimensional (3D) lane information using the offset information.