Railway Image Monitoring for Predictive Obstruction Detection

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current railway condition monitoring is reactive and based on empirical data, lacking effective use of artificial intelligence to analyze large volumes of video data, which is computationally intensive and inefficient, leading to insufficient prevention of trespassing-related casualties and traffic delays at grade crossings.

Innovation Solution

Implementing machine learning-enhanced systems that utilize imaging devices and neural networks to recognize objects and conditions, predict future railway conditions, and generate alerts for train operators, integrating computer vision techniques and custom-trained neural networks for real-time monitoring and assessment of railway conditions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If machine learning techniques are used to analyze large volumes of video data, then detection accuracy and predictive capability are improved, but computation resource requirements increase significantly

Engineering Contradiction:

Improvedetection accuracyVSAvoidcomputation resource requirements

Core Design Contradiction:

ReliabilityVSUse of energy by moving object

Solution Approach 1:

The system segments the video data processing into multiple stages: initial frame extraction, object detection, condition assessment, and predictive analysis. By processing data in segmented portions rather than analyzing entire video streams simultaneously, the system reduces computational resource requirements while maintaining detection accuracy.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system performs preliminary actions by pre-processing video data to extract key frames and identify potential objects before performing comprehensive analysis. This preliminary screening reduces the volume of data requiring intensive machine learning processing, thereby lowering computation resource requirements while maintaining reliability.

Inventive Principle:

Principle #10Preliminary action

2Device complexity

If reactive monitoring based on empirical data is used, then system complexity is reduced, but prevention capability and operational efficiency deteriorate

Engineering Contradiction:

Improvesystem complexityVSAvoidprevention capability

Core Design Contradiction:

Device complexityVSReliability

Solution Approach 1:

The system implements feedback mechanisms where detected conditions and predicted outcomes are fed back into the monitoring system to continuously improve detection algorithms and predictive models. This feedback loop enables the system to learn from historical data and improve prevention capability without requiring excessive complexity.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The system replaces traditional mechanical monitoring approaches with machine learning-based automated detection and prediction. This substitution enables more effective prevention capability while managing complexity through software-based solutions rather than hardware-intensive systems.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Measurement precision

If comprehensive video data analysis is performed, then detection of obstructions and conditions is improved, but processing time and computational efficiency deteriorate

Engineering Contradiction:

Improvecondition detection precisionVSAvoidprocessing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system applies local quality analysis by focusing computational resources on specific regions of interest within the video data, such as track areas and potential obstruction zones, rather than uniformly processing entire frames. This targeted approach maintains detection precision for critical conditions while reducing overall processing time.

Inventive Principle:

Principle #3Local quality

Solution Approach 2:

The system performs partial analysis by processing only the most critical video frames and data portions necessary for detecting obstructions and conditions. This selective processing approach maintains adequate measurement precision for safety-critical detections while significantly reducing processing time compared to comprehensive analysis.

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS12479482B2Systems and methods for machine learning enhanced railway condition monitoring, assessment and prediction

Publication Date: 2025.11.25 RUTGERS THE STATE UNIV

AI summary

Systems and methods enable automated intelligent railway monitoring using railway image data from an imaging device. The railway image data includes image frames produced by the imaging device. A railway object recognition model is used to identify objects within the image frames and a railway condition is determined based on the object. A location and time associated with the railway condition are determined based on railway image data. Railway metrics are generated based on the railway condition and historical railway conditions. A railway condition prediction model is used to predict future railway conditions affecting an operating condition, an operating demand or both based on the railway metrics. An operating recommendation is determined to address the future railway condition and an alert is generated to with an indication of railway condition, the operating recommendation or both.