Intersection Collision Prediction Using Yielding Intention Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods struggle to accurately predict collision risks at intersections due to unpredicted conditions and uncertainties, leading to either overly cautious safety measures that hinder traffic flow or insufficient safety measures that result in accidents.
Innovation Solution
A system utilizing a camera and radar to predict the traveling intentions of crossing vehicles, determine collision risk by comparing time to collision with minimum and maximum arrival times, and adjust vehicle operations based on the driver's yielding intention using an intelligent driver model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the degree of collision risk is conservatively determined to ensure safety, then safety is improved, but traffic flow is hindered
Solution Approach 1:
The system changes the parameter of risk assessment from static TTC-based values to dynamic values that incorporate predicted traveling paths and yielding intentions. By adjusting these parameters based on real-time analysis of crossing vehicle behavior, the system achieves both high safety and smooth traffic flow without conservative over-restriction
Solution Approach 2:
The system performs preliminary prediction of crossing vehicle traveling paths and yielding intentions before the actual collision risk materializes. By anticipating the crossing vehicle's future actions and preparing appropriate responses in advance, the system prevents unnecessary traffic disruptions while maintaining safety
2Ease of operation
If conventional TTC-based risk determination is used, then the system is simple to operate, but it cannot accurately predict collision risks in unpredictable intersection conditions
Solution Approach 1:
The system segments the complex task of collision risk assessment into distinct modules: path prediction, yielding intention recognition, and risk determination. Each module handles a specific aspect independently, maintaining system simplicity while improving overall accuracy through specialized analysis of each component
Solution Approach 2:
The system introduces an intermediary layer of path prediction and yielding intention analysis between the raw sensor data and the final risk determination. This intermediary processing layer translates complex unpredictable behaviors into structured risk assessments, improving accuracy without significantly increasing operational complexity
Data Source
AI summary
Disclosed herein is a system includes a camera having a field of view around a host vehicle and configured to output image data, a radar having a sensing area around the host vehicle and configured to output radar data, and a controller electrically connected to the camera and the radar. The controller may determine a future traveling path of the host vehicle based on a preceding direction of the host vehicle from the image data and/or the radar data, determine a future traveling path of a crossing vehicle based on a preceding direction of the crossing vehicle moving in a direction crossing the preceding direction of the host vehicle, determine a point at which the future traveling path of the host vehicle crosses the future traveling path of the crossing vehicle as a predicted collision position, determine whether yielding intention of a driver of the crossing vehicle is present when the crossing vehicle approaches the predicted collision position, and avoid a collision with the crossing vehicle based on the yielding intention of the driver of the crossing vehicle.