Connected Vehicle Speed Anomaly Detection via Road Segment Profiles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current traffic management and monitoring systems rely on hardware-based detectors and user reports, which are limited in their ability to detect speed anomalies and near misses, and do not effectively identify areas of reckless driving or road hazards.
Innovation Solution
A system and method using roadside units and onboard units to collect and analyze vehicle traffic data, employing machine learning techniques like k-means clustering and Frechet distance to detect speed anomalies by comparing vehicle profiles with predefined thresholds, reducing the need for dedicated detection infrastructure.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If hardware-based detectors (magnetometers, microwave radar, inductive loops, cameras) are used to detect traffic incidents and queues, then detection reliability is improved, but device complexity and infrastructure cost increase
Solution Approach 1:
The patent uses virtual copies of physical detectors by processing video images through image processing algorithms to create virtual detection zones. Instead of deploying physical magnetometers or radar devices, the system creates software-based detectors that replicate detection functionality through computer vision, thereby reducing hardware complexity while maintaining detection capability
Solution Approach 2:
The patent replaces mechanical and electromagnetic detection systems (magnetometers, inductive loops, radar) with an optical-computational system. Video cameras capture visual information, and image processing algorithms analyze the captured frames to detect vehicles, queues, and incidents, substituting physical sensing mechanisms with optical capture and computational analysis
2Device complexity
If user reports via mobile applications are used to detect incidents, then infrastructure cost is reduced, but measurement precision and response time worsen due to reliance on active user reporting
Solution Approach 1:
The system enables self-service detection by automatically analyzing video feeds from existing surveillance cameras without requiring manual user intervention. The image processing algorithms autonomously detect vehicles, determine their positions, identify queues, and report incidents, making the detection system self-operating and eliminating the need for active user reporting while improving detection accuracy
Solution Approach 2:
The patent implements continuous detection by processing video frames in real-time as they are captured. Unlike periodic user reports, the system continuously monitors the detection zones, analyzing each video frame to update vehicle positions and detect incidents immediately, ensuring uninterrupted and timely detection of traffic conditions
3Measurement precision
If comprehensive vehicle traffic data is collected and analyzed using machine learning, then speed anomaly detection precision is improved, but loss of information and privacy concerns increase
Solution Approach 1:
The patent extracts only the specific information needed for detection purposes from the video data. The image processing algorithms identify vehicle positions, speeds, and patterns without capturing or storing identifiable information about drivers or passengers. By extracting only relevant traffic flow data and discarding personal identifiers, the system achieves precise speed anomaly detection while preserving privacy
Solution Approach 2:
The system uses video images as an intermediary medium that captures traffic information without directly observing or identifying individuals. The image processing algorithms analyze the visual data to detect vehicle speeds and patterns, using the video feed as a mediator that provides necessary traffic information while maintaining a barrier that protects driver and passenger privacy
Data Source
AI summary
A system for detecting speed anomalies in a connected vehicle infrastructure environment includes a plurality of roadside units configured to collect vehicle traffic data from vehicles travelling in a road network; a vehicle traffic data evaluation module configured to receive collected vehicle traffic data from the plurality of roadside units, the vehicle traffic data comprising multiple data sets of multiple vehicles, extract first data from the collected vehicle traffic data, combine and/or conform the first data with second data, create vehicle profiles based on combined and/or conformed first data and second data and road segment profiles based on clustering and filtering of the vehicle profiles, compare a first vehicle profile of the collected vehicle traffic data with a road segment profile, and detect and output an anomaly of the first vehicle profile when the first vehicle profile is outside a predefined threshold.