Single-Camera Mobile Robot 3D Mapping Using Particle Re-Projection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional mobile robots rely on 1-dimensional or 2-dimensional sensors to recognize surroundings, which limits their ability to accurately detect and navigate through complex environments, failing to effectively utilize 3D information for autonomous cleaning.

Innovation Solution

A mobile robot equipped with a single camera that re-projects particles from one image frame to another based on matching feature points, allowing for the precise extraction of 3D information by generating particles on a virtual line passing through feature points and the camera's center, enabling the creation of a 3D map of its surroundings.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If a single camera is used for 3D recognition, then device complexity is reduced, but measurement precision of 3D information deteriorates

Engineering Contradiction:
Improvenumber of sensorsVSAvoid3D information extraction accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent transitions from 2D image plane coordinates to 3D spatial coordinates by introducing depth information through particle re-projection. Particles are generated on a virtual line extending from the camera center through feature points, and their positions are refined by projecting onto subsequent image frames, effectively adding the depth dimension using only a single camera.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

Solution Approach 2:

The patent introduces particles as intermediary elements between the camera and the feature points. These particles serve as virtual probes that are re-projected across multiple frames to infer depth information, acting as a mediator that enables 3D reconstruction without requiring multiple cameras or complex sensor arrays.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Device complexity

If conventional 2D sensors are used for environment recognition, then device complexity is low, but the ability to detect and navigate complex environments deteriorates

Engineering Contradiction:
Improvesensor configurationVSAvoidenvironmental recognition capability
Core Design Contradiction:
Device complexityVSAdaptability or versatility

Solution Approach 1:

The patent enhances environmental recognition capability by extracting 3D spatial information from 2D image sequences. By re-projecting particles onto subsequent frames and calculating depth based on feature point matching, the system gains three-dimensional awareness while maintaining the simplicity of a single camera setup.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

Solution Approach 2:

The patent replaces complex multi-sensor mechanical systems with a computational approach using a single camera. Instead of using multiple cameras or depth sensors, the system uses image processing algorithms and particle re-projection techniques to achieve 3D environmental understanding, substituting mechanical complexity with computational intelligence.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentEP2413772B1Mobile robot with single camera and method for recognizing 3D surroundings of the same
Publication Date: 2019.05.08 LG ELECTRONICS INC
  • EP2413772B1 patent drawingFigure 1
  • EP2413772B1 patent drawingFigure 2
  • EP2413772B1 patent drawingFigure 3

AI summary

Disclosed are a mobile robot with a single camera capable of performing a cleaning process with respect to surroundings, and capable of more precisely making a 3D map of the surroundings including a plurality of feature points, and a method for recognizing 3D surroundings of the same. According to the method, images of the surroundings are captured, and a preset number of particles with respect to feature points of a first image are projected to a second image based on matching information of feature points extracted from the two images sequentially captured, thereby extracting 3D information of the surroundings.