Floor Pattern Code Maps for Robot Location and Orientation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous driving robots face challenges in accurately determining their location and orientation in spaces with repetitive patterns, such as long corridors, due to the inefficiencies in data size and computation requirements of existing mapping methods, and errors in LiDAR-based SLAM and image-based systems.
Innovation Solution
A system and method for generating a code block map using floor patterns, which involves extracting unit patterns from camera images, assigning identification codes, and generating a code block map to reduce computation, and recognizing location and orientation by comparing a floor pattern code string with the code block map.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If LiDAR-based SLAM or image-based mapping methods are used to determine robot location and orientation, then the robot can navigate in various environments, but the computation load increases and accuracy decreases in spaces with repetitive patterns such as long corridors
Solution Approach 1:
The patent extracts only the essential features from floor patterns (geometric shapes, colors, textures of tiles) to create simplified code blocks, removing redundant visual information. This extraction process reduces computation load while maintaining sufficient accuracy for location recognition in repetitive pattern spaces
Solution Approach 2:
The patent transforms continuous image data into discrete code blocks by changing the parameter representation from pixel-level continuous values to block-level discrete codes. This parameter transformation simplifies the data structure and reduces computation requirements while preserving location information
2Loss of information
If detailed image information is used for mapping, then the map contains comprehensive environment information, but the data size increases and requires large computation for image comparison
Solution Approach 1:
The patent segments the continuous floor pattern image into discrete code blocks representing different tile types. Each block is assigned a unique code based on its pattern characteristics, transforming large image data into compact coded representations that preserve essential environmental information while reducing data size
Solution Approach 2:
The patent creates simplified code block representations that copy only the essential identifying features of floor patterns rather than storing complete image data. These code blocks serve as compressed copies that enable location recognition without requiring full image storage and comparison
Data Source
AI summary
Provided is a technology for recognizing a location or orientation of a robot using patterns of a floor on which the robot travels, and the technology includes two interrelated sub-technologies. The first sub-technology is a technology for generating a code block map based on floor pattern recognition, in which a code block map is generated using patterns of a floor on which a robot travels, and the second sub-technology is a technology for recognizing a location and orientation of a robot based on a floor pattern code string, in which a location and orientation of a traveling robot are recognized using a generated code block map.