Mobile Robot Key Frame Update via Stereo SLAM

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current mobile robots using dead reckoning navigation with monocular cameras or laser scanners face significant position recognition errors due to limitations in distance measurement and high costs of laser scanners, necessitating the development of more accurate and cost-effective methods for simultaneous localization and mapping (SLAM) using stereo cameras.

Innovation Solution

An apparatus and method for updating key frames of a mobile robot using approximated difference of Gaussian (ADoG) based feature points, which estimates position through odometry information from stereo imaging and inertial data, and performs dense stereo alignment to correct positional errors and detect obstacle distances without dedicated sensors.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If monocular camera is used for navigation, then device cost is reduced, but position recognition accuracy deteriorates due to inability to measure distance

Engineering Contradiction:

Improvedevice costVSAvoidposition recognition accuracy

Core Design Contradiction:

Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent combines monocular camera imaging with inertial measurement unit (IMU) data to create a fused navigation system. The camera provides visual features while the IMU provides motion information, and their fusion compensates for the monocular camera's inability to measure depth, thereby maintaining low cost while improving position recognition accuracy.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The patent introduces visual odometry as an intermediary method that estimates depth and position by analyzing optical flow and feature point movements across multiple camera frames. This intermediary computation layer translates 2D image data into 3D position information, enabling accurate navigation without expensive laser scanners.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Measurement precision

If laser scanner is used for navigation, then position recognition accuracy is improved, but device cost increases significantly

Engineering Contradiction:

Improveposition recognition accuracyVSAvoiddevice cost

Core Design Contradiction:

Measurement precisionVSEase of manufacture

Solution Approach 1:

The patent replaces expensive laser scanners with inexpensive monocular cameras that capture multiple sequential images. By processing these short-lived image sequences through visual odometry algorithms, the system achieves laser-scanner-level accuracy at a fraction of the cost, effectively using cheap disposable image data instead of expensive dedicated ranging sensors.

Inventive Principle:

Principle #27Cheap short-living objects (Disposable)

Solution Approach 2:

The patent substitutes the mechanical laser scanning system with an optical imaging system (monocular camera) combined with computational processing. Instead of using active laser ranging, the system passively captures images and computationally derives position and depth information, replacing complex mechanical scanning with simpler optical capture and software processing.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Device complexity

If dead reckoning navigation is used, then device complexity is reduced, but accumulated error increases over time

Engineering Contradiction:

Improvenavigation system complexityVSAvoidposition accuracy over time

Core Design Contradiction:

Device complexityVSReliability

Solution Approach 1:

The patent implements visual feedback by continuously comparing observed feature point movements with predicted movements from dead reckoning. When discrepancies are detected, the system corrects accumulated errors by adjusting the position estimate based on actual visual observations, creating a closed-loop navigation system that maintains accuracy over time.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The patent makes the monocular camera serve multiple functions: it captures images for visual odometry, extracts feature points for position estimation, and provides feedback for error correction. This multi-functional use of a single sensor eliminates the need for separate correction sensors, maintaining system simplicity while improving long-term reliability.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

Improves position recognition accuracy by using stereo camera-based SLAM to correct odometry errors and detect obstacles, reducing the need for expensive sensors and enhancing mobility and service capabilities of mobile robots.

Implementation Method 1

utilizing the parallax value of the seeds projected from the previously acquired key frames

Methodology Applied

Scientific EffectParallax: Parallax

Data Source

PatentUS10133279B2Apparatus of updating key frame of mobile robot and method thereof

Publication Date: 2018.11.20 YUJIN ROBOT

AI summary

Disclosed are an apparatus and a method of updating a key frame of a mobile robot. An apparatus for updating a key frame of a mobile robot includes a key frame initializing unit which initializes seeds which constitute a key frame in a first position of the mobile robot and a key frame updating unit which projects seeds in the initialized key frame onto a first image photographed in the first position to obtain a coordinate according to each of the seeds and projects the seeds with the coordinates onto a second image photographed in a second position in accordance with movement of a mobile robot to update the seeds of the key frame as a projected result.