Vehicle Collision Risk Prediction Using Virtual Lane Distances
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The high construction cost and time required for detailed maps in autonomous driving systems make it challenging to implement accurate nationwide navigation for autonomous vehicles, which affects the calculation of initial paths and collision risk prediction.
Innovation Solution
A vehicle system that calculates distances between itself and surrounding vehicles by using full width information, virtual lines, and virtual points, applying weights based on speed and road curvature to predict collision risks, and a method to control the vehicle accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If detailed maps are constructed for autonomous driving navigation, then navigation accuracy is improved, but construction cost and time increase significantly
Solution Approach 1:
The patent uses virtual lines and virtual points as simplified copies of actual road boundaries and vehicle positions. Instead of requiring detailed real-world map data, the system creates virtual representations that capture essential geometric relationships for collision risk assessment, achieving accurate navigation with reduced map construction requirements
Solution Approach 2:
The patent introduces virtual lines as intermediary elements between the vehicle and actual road boundaries. These virtual lines serve as mediators that simplify the complex task of detailed map construction by providing a computationally efficient representation of road geometry that maintains navigation accuracy
2Measurement precision
If traditional distance calculation methods are used between vehicle and surrounding vehicles, then calculation simplicity is maintained, but collision risk prediction accuracy deteriorates
Solution Approach 1:
The patent segments the distance calculation process into multiple components: distance from vehicle to virtual line, distance from surrounding vehicle to its virtual line, and relative position adjustments. This segmentation allows each component to be calculated independently and accurately, improving overall collision risk prediction while maintaining computational efficiency
Solution Approach 2:
The patent transitions from calculating simple one-dimensional distance between vehicles to a multi-dimensional approach that considers distances to virtual lines, lane positions, and relative orientations. This dimensional expansion enables more accurate collision risk assessment by capturing the spatial relationship between vehicles and road geometry
Data Source
AI summary
A vehicle for predicting a risk of collision includes a controller configured to: calculate distances between the vehicle and left and right lines of a first lane, respectively, using a position of the vehicle and first lane width information of the first lane, calculate distances between the surrounding vehicle and left and right lines of a second lane, respectively, using a position of the surrounding vehicle and second lane width information of the second lane, calculate a second distance between the vehicle and the surrounding vehicle by reflecting the calculated distances between the vehicle and the left and right lines of the first lane or the calculated distances between the surrounding vehicle and the left and right lines of the second lane to a first distance, and predict a risk of collision between the vehicle and the surrounding vehicle based on the second distance.