Vehicle Position Estimation Using Type-Based Feature Point Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle position estimation systems face accuracy issues due to erroneous matching between detection data and map data, particularly when objects with similar shapes are present, leading to decreased estimation accuracy.
Innovation Solution
A position estimation apparatus that extracts feature points from camera images, recognizes object types, and searches for corresponding points in environmental maps using type information, employing bundle adjustment with constraint conditions to minimize estimation errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If feature point matching is performed without type information, then the matching process is simple and fast, but erroneous matching occurs when objects with similar shapes are present, decreasing estimation accuracy
Solution Approach 1:
The patent introduces type information as an intermediary element that mediates between feature points and their matching process. By categorizing objects into types (e.g., division lines, crosswalks, buildings) and using this type information as a filtering criterion, the system prevents erroneous matching between similar-shaped objects while maintaining a relatively simple matching framework. The type information acts as a mediator that guides the matching process without requiring complete redesign of the matching algorithm.
2Measurement precision
If type information is added to map data for every object, then matching accuracy improves by preventing erroneous matching, but the data storage requirement and processing complexity increase
Solution Approach 1:
The patent applies local quality by assigning type information selectively to specific map objects based on their characteristics. Rather than uniformly adding complex data structures to all map elements, the system categorizes objects into types (such as division lines, crosswalks, traffic lights, buildings) and stores type information only where needed. This localized approach improves matching accuracy for critical objects while minimizing the overall increase in data storage requirements.
3Productivity
If conventional matching methods are used, then the processing speed is fast, but erroneous matching occurs leading to position estimation errors
Solution Approach 1:
The patent implements preliminary action by pre-categorizing map objects into types and storing this type information in advance in the map data structure. Before the actual feature point matching process begins, the system has already organized the map data with type classifications, allowing for rapid filtering and verification during matching. This preliminary organization enables the system to maintain fast processing speeds while significantly improving matching reliability through type-based validation.
Data Source
AI summary
A position estimation apparatus estimates a position of a vehicle based on a first feature point of an object included in a camera image and a second feature point of the object included in map information. The apparatus includes a memory storing type information indicating a type of the object corresponding to the second feature point together with the map information, and a microprocessor performing: extracting the first feature point from the camera image, recognizing the type of the object corresponding to the first feature point based on the camera image, searching for the second feature point corresponding the first feature point from the map information based on the first feature point, the type of the object corresponding to the first feature point, and the type information, and estimating the position of the vehicle based on the first feature point and the second feature point.