Visual SLAM Position Estimation with GNSS Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing visual-SLAM techniques face challenges in accurately estimating camera position and orientation due to scale drift and limited correction capabilities, especially when applied to ordinary vehicle videos, leading to inaccurate mapping and positioning for services like automatic driving.
Innovation Solution
A position estimation system that incorporates GNSS information for optimization and correction of key frames and feature points, using a two-stage processing approach to minimize scale drift and improve estimation accuracy without requiring loop closures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If visual-SLAM technique is used to estimate camera position and orientation, then positioning can be performed without external infrastructure, but scale drift occurs leading to reduced measurement precision
Solution Approach 1:
The patent introduces an environmental map as an intermediary data structure that stores three-dimensional positions of feature points and camera positions. This map serves as a mediator between multiple video sequences, enabling cross-sequence position estimation and reducing scale drift by providing a stable reference framework that accumulates positioning information over time.
Solution Approach 2:
The patent creates a virtual copy of the physical environment through the environmental map, which replicates the spatial relationships and feature point positions. This virtual model allows the system to estimate camera positions by matching observed feature points against the stored map data, thereby achieving accurate positioning without direct continuous measurement.
2Adaptability or versatility
If ordinary vehicle videos are used for position estimation, then the system can work with readily available data, but the accuracy is insufficient for services like automatic driving
Solution Approach 1:
The patent enables the system to create and utilize its own environmental map from the video data it processes. The environmental map is built by detecting feature points in video frames and estimating camera positions, then storing this information for future position estimation. This self-generated map allows the system to continuously improve its positioning accuracy using the same ordinary vehicle videos without requiring external infrastructure or specialized equipment.
3Measurement precision
If GNSS information is used for correction, then position accuracy can be improved, but the system becomes dependent on external satellite signals
Solution Approach 1:
The patent performs preliminary construction of the environmental map by detecting feature points and estimating camera positions from video frames before actual position estimation is needed. This pre-built map contains stored three-dimensional positions of feature points and camera positions that can be used for rapid and accurate position estimation without requiring real-time GNSS signals or complex calculations during operation.
Data Source
AI summary
A position estimation system includes one or more memories and one or more processors configured to acquire a first imaging position measured at a time of imaging a first image among a plurality of images imaged in time series, perform, based on a feature of the first image, calculation of a second imaging position of the first image, and perform, in accordance with a constraint condition that reduces a deviation between the first imaging position and the second imaging position, correction of at least one of the second imaging position or a three-dimensional position of a point included in the first image calculated based on the feature of the first image.