3D Visual Feature Maps for Low-Cost Vehicle Localization

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing autonomous vehicle technologies rely heavily on expensive sensors like LiDAR, RADAR, and high-precision GPS, which hinder the widespread adoption of autonomous vehicles due to cost constraints.

Innovation Solution

A method and system for generating a visual feature map using image processing to estimate vehicle location and posture, utilizing a camera to detect keypoints, extract visual feature descriptors, and associate them with 3D coordinate values and image capturing direction, enabling accurate location estimation without expensive sensors.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If expensive sensors like LiDAR, RADAR, and high-precision GPS are used, then measurement precision of vehicle location is improved, but device cost increases

Engineering Contradiction:

Improvevehicle location estimation accuracyVSAvoidsensor cost

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces expensive, long-lasting sensors (LiDAR, high-precision GPS) with cheaper alternatives (standard cameras, consumer-grade sensors) that can be used disposable-style in terms of cost consideration, achieving similar localization accuracy through software processing rather than hardware superiority

Inventive Principle:

Principle #27Cheap short-living objects (Disposable)

Solution Approach 2:

The patent substitutes mechanical/optical sensing systems (LiDAR, RADAR) with computational image processing systems. Instead of using mechanical sensors to directly measure distance and position, the system uses camera images processed through keypoint detection, descriptor extraction, and feature matching algorithms to achieve location estimation

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 3:

The patent changes the fundamental parameters of the sensing approach by transitioning from direct physical measurement (time-of-flight, electromagnetic wave reflection) to visual feature-based indirect measurement. This involves changing from measuring distance directly to inferring position from visual feature correspondence between images

Inventive Principle:

Principle #35Parameter changes

2Measurement precision

If visual feature maps with detailed geometric information are stored, then location estimation accuracy is improved, but memory usage increases

Engineering Contradiction:

Improvelocation estimation accuracyVSAvoidmemory storage capacity

Core Design Contradiction:

Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent extracts only the essential visual features (keypoints and their descriptors) from the complete scene geometry, storing only these extracted features rather than the full 3D point cloud or complete image data. This extraction process removes redundant information while preserving location-estimation capability

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent segments the visual scene into discrete keypoint features rather than storing continuous image or point cloud data. Each keypoint is independently detected, described, and stored with its 3D coordinates, creating a segmented representation that is more storage-efficient while maintaining estimation accuracy

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS12450772B2Method and system for generating visual feature map

Publication Date: 2025.10.21 NAVER CORP

AI summary

A method for generating a visual feature map includes receiving a first image, detecting a first set of keypoints from the first image, extracting a visual feature descriptor of each of the first set of keypoints, receiving a first set of 3D point cloud data associated with a location where the first image is captured, and determining a 3D coordinate value of each of the first set of keypoints using the first set of 3D point cloud data.