Lane Crossing Detection via Neural Network Self-Calibration

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current lane departure warning systems require lengthy and error-prone calibration processes to determine camera positioning, leading to inaccurate vehicle lane drift calculations and inefficient resource utilization.

Innovation Solution

A video system that detects lane line crossings and classifies lane changes using forward-facing video data processed with neural network models, without requiring explicit camera positioning or manual calibration.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual calibration is performed to determine camera positioning, then measurement precision of lane drift is improved, but loss of time and productivity deteriorate due to lengthy calibration process

Engineering Contradiction:

Improvelane drift detection accuracyVSAvoidcalibration time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system performs self-calibration by automatically determining camera positioning parameters through processing forward-facing video data to generate histograms of lane line positions, fitting probability density functions, and calculating mean and standard deviation values without requiring manual intervention or external calibration tools

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent replaces the manual mechanical calibration process with an automated computational system that uses neural networks to process video data and automatically determine camera positioning parameters, eliminating the need for physical calibration equipment and manual adjustment

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Manufacturing precision

If manual calibration is performed to determine camera positioning, then manufacturing precision of lane drift calculation is improved, but device complexity increases due to calibration requirements

Engineering Contradiction:

Improvelane drift calculation accuracyVSAvoidcalibration process complexity

Core Design Contradiction:

Manufacturing precisionVSDevice complexity

Solution Approach 1:

The system automatically determines camera positioning parameters through self-calibration by processing forward-facing video data to generate histograms of lane line positions, fitting probability density functions, and calculating mean and standard deviation values without requiring manual intervention or external calibration tools

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent transforms the calibration process from a manual parameter-setting operation into an automated parameter-extraction process by using neural networks to analyze video data and automatically determine camera positioning parameters based on statistical analysis of lane line positions

Inventive Principle:

Principle #35Parameter changes

3Measurement precision

If complex calibration processes are used to determine camera positioning, then measurement precision of lane line crossing detection is improved, but ease of operation deteriorates

Engineering Contradiction:

Improvelane line crossing detection accuracyVSAvoidsystem setup ease

Core Design Contradiction:

Measurement precisionVSEase of operation

Solution Approach 1:

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent performs automatic camera positioning determination as a preliminary step before lane drift detection, using neural networks to process video data and establish calibration parameters in advance, so that the system is ready for operation without requiring subsequent manual calibration

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS20250131741A1Systems and methods for detecting lane crossings and classifying lane changes

Publication Date: 2025.04.24 VERIZON PATENT & LICENSING INC

AI summary

A device may receive forward facing video data associated with a vehicle, and may process the forward facing video data, with neural network models, to detect lane lines and to determine classifications for the lane lines. The device may utilize the forward facing video data to generate a histogram of horizontal positions of the vehicle, and may fit probability density functions on the histogram to calculate a mean and a standard deviation. The device may utilize the mean and the standard deviation to identify a crossing interval, and may classify the forward facing video data as a lane crossing or a lane change based on the crossing interval. The device may calculate a lane crossing score or may calculate a lane change score. The device may perform actions based on the lane crossing score or the lane change score.