Apparatus and method of generating map data of cleaning space

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Robotic cleaning apparatuses face challenges in accurately generating map data of cleaning spaces and identifying objects within these spaces, which is crucial for effective navigation and cleaning operations.

Innovation Solution

The apparatus employs AI models, including neural networks, to generate basic map data, recognize objects, and divide the space into regions, determining identification values for these regions based on object information, enabling precise mapping and navigation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If rule-based smart systems are used for cleaning space mapping, then device complexity is reduced, but measurement precision and object recognition accuracy deteriorate

Engineering Contradiction:
Improvesystem complexityVSAvoidobject recognition accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent replaces rule-based mechanical systems with deep learning-based AI systems for object recognition and mapping. The robotic cleaning apparatus uses neural networks to automatically learn and recognize objects, spaces, and patterns without predefined rules, significantly improving measurement precision and object recognition accuracy while accepting increased computational complexity.

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

Solution Approach 2:

The patent transforms the operational parameters from simple rule-based logic to complex AI model parameters including neural network weights, learning rates, and deep learning epochs. This parameter transformation enables the system to adapt to various cleaning environments and improve recognition accuracy through continuous learning and optimization.

Inventive Principle:
Principle #35Parameter changes

2Productivity

If basic map data generation is used without object recognition, then productivity is improved by simplifying processing, but measurement precision and spatial understanding deteriorate

Engineering Contradiction:
Improvemapping efficiencyVSAvoidspatial understanding accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent implements preliminary object recognition and classification during the mapping process itself, rather than as a separate subsequent step. The AI system identifies objects, determines their types, and integrates this information into the map data structure in real-time, enabling both efficient productivity and precise spatial understanding simultaneously.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent merges the object recognition function with the map generation function into a unified AI processing system. The neural network simultaneously performs both tasks by integrating object detection outputs directly into the spatial mapping data structure, achieving synergistic effects where both productivity and precision are improved.

Inventive Principle:
Principle #5Merging (Combining)

3Measurement precision

If AI models are used for object recognition and region identification, then measurement precision and object recognition accuracy are improved, but device complexity and computational requirements increase

Engineering Contradiction:
Improveobject recognition accuracyVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the AI processing into distinct functional modules: object detection module, object classification module, region identification module, and map generation module. Each module handles specific computational tasks, allowing the system to manage complexity through modular design while maintaining high measurement precision through specialized processing for each function.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentUS11334084B2Apparatus and method of generating map data of cleaning space
Publication Date: 2022.05.17 SAMSUNG ELECTRONICS CO LTD
  • US11334084B2 patent drawing
  • US11334084B2 patent drawing
  • US11334084B2 patent drawing

AI summary

Provided are an artificial intelligence (AI) system using a machine learning algorithm like deep learning, or the like, and an application thereof. A robotic cleaning apparatus that generates map data includes a communication interface comprising communication circuitry, a memory configured to store one or more instructions, and a processor configured to control the robotic cleaning apparatus by executing the one or more instructions. The processor is configured, by executing the one or more instructions, to control the robotic cleaning apparatus to: generate basic map data related to a cleaning space, and generate object information regarding at least one object in the cleaning space, the object information being generated based on information obtained by the robotic cleaning apparatus regarding the object in a plurality of different positions of the cleaning space, and including information about a type and a position of the object.