Vehicle Positioning Using Visual Map Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current vehicle positioning systems, such as GPS, are not accurate enough to safely navigate vehicles within intersections, where a significant percentage of fatal accidents occur, due to limitations in determining the vehicle's location relative to map features.
Innovation Solution
A method that captures images of the scene, identifies points corresponding to features, and compares them to a map to determine the vehicle's position, using image processing techniques like edge detection and predictive filters to improve accuracy and reliability, allowing for precise location and motion estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If GPS positioning system is used to determine vehicle location, then the system is simple and provides coverage over large areas, but the positioning accuracy is insufficient (error of tens of metres) for intersection safety requirements
Solution Approach 1:
The patent combines GPS positioning with visual recognition technology. The system integrates a camera to capture images of road features, identifies these features through image processing, and merges this visual information with GPS data to achieve precise positioning within intersections, thereby resolving the accuracy limitation of GPS alone.
Solution Approach 2:
The patent introduces map data as an intermediary element. The system compares identified road features from captured images against pre-stored map data of intersection features. This intermediary map matching process enables the system to achieve high positioning accuracy by determining the vehicle's location based on the spatial relationship between detected features and map features.
2Measurement precision
If image processing techniques are used to identify features and determine position, then positioning accuracy improves to 10-20cm, but processing time and computational complexity increase
Solution Approach 1:
The patent pre-stores detailed map data of intersection features before the vehicle arrives. By having the map data ready in advance, the system can quickly compare captured images against known features without performing complex real-time analysis of all image data, thus reducing processing time while maintaining high accuracy.
Solution Approach 2:
The patent focuses image processing resources on specific regions of interest within the captured images, particularly areas where intersection features are likely to appear. Rather than processing the entire image uniformly, the system concentrates computational effort on detecting and identifying critical road features, thereby reducing overall processing time while maintaining positioning accuracy.
Data Source
Figure 1
Figure 2~3
Figure 4
AI summary
A method of determining the location of a vehicle (100) on a map, the map comprising a plurality of points associated with features, comprising: capturing (202), typically using a video camera (102) an image of a scene from the vehicle (100), identifying (204) points in the image corresponding to features in the scene, and comparing (208) the points in the captured image to the map to determine the position of the vehicle. The method typically further comprises the step of capturing (216) at least one further image of the scene, identifying points (218) in the at least one further image and comparing (220) the points identified in the image and the or each further images. The method may also comprise comprises modeling (222) the motion of the vehicle (100) using a predictive filter, such as a Kalman filter, or an Extended Kalman filter.