Cleaning Robot AI-Driven Guard Control for Object Manipulation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional cleaning robots are limited by their sensor combinations, which restrict their ability to obtain specific information about objects in their vicinity, leading to repetitive pattern-based obstacle avoidance without adapting to different objects, necessitating a method to determine optimal tasks based on identified objects.

Innovation Solution

A cleaning robot equipped with a camera, processor, and a guard portion that captures images of objects and applies them to trained AI models to determine tasks, controlling the guard portion to descend and move objects, allowing for tailored actions such as avoidance or relocation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a general cleaning robot uses only basic sensors for obstacle detection, then the device complexity is low, but the measurement precision of object information is insufficient, leading to repetitive pattern-based avoidance

Engineering Contradiction:
Improveobject information accuracyVSAvoidsensor combination complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The cleaning robot segments object recognition into multiple stages: initial obstacle detection by basic sensors, detailed image capture by camera, and AI-based classification. This segmentation allows the system to achieve high measurement precision for object information while keeping the overall device complexity manageable by only activating advanced sensors when needed.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The camera acts as an intermediary between the basic sensors and the AI processing system. When basic sensors detect an obstacle, the camera captures detailed images that serve as intermediate data, enabling the AI model to precisely classify the object type without requiring all advanced sensors to operate continuously, thus balancing measurement precision and device complexity.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Productivity

If the cleaning robot performs only avoidance driving based on limited sensor data, then the ease of operation is maintained, but the productivity is reduced due to inability to clean obstructed areas

Engineering Contradiction:
Improvecleaning area coverageVSAvoidtask decision complexity
Core Design Contradiction:
ProductivityVSEase of operation

Solution Approach 1:

The cleaning robot uses AI-based object classification to automatically determine the appropriate task without human intervention. The system self-services by classifying objects (e.g., distinguishing between movable items like shoes and immovable structures like vents) and autonomously selecting whether to avoid, push, or clean around them, thereby increasing productivity while maintaining ease of operation.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The robot changes its operational parameters based on AI-classified object attributes. When an object is identified as movable (low height, lightweight), the robot changes from avoidance mode to pushing mode. This dynamic parameter adjustment based on object characteristics enables the robot to clean previously obstructed areas, significantly improving productivity without requiring complex manual task programming.

Inventive Principle:
Principle #35Parameter changes

3Adaptability or versatility

If the cleaning robot uses AI model to determine optimal tasks based on object identification, then the adaptability to different objects is improved, but the loss of time for image processing and analysis increases

Engineering Contradiction:
Improvetask differentiation capabilityVSAvoidobject recognition processing time
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

The system performs preliminary action by pre-training AI models with extensive object data before deployment. During operation, the pre-trained model rapidly classifies objects without requiring time-consuming analysis. The robot also performs preliminary object detection with basic sensors before activating the camera, so AI processing only begins when necessary, reducing overall processing time while maintaining high adaptability.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The cleaning robot implements periodic action by capturing images at specific intervals or when triggered by basic sensor detection rather than continuously. The AI model processes images periodically based on these triggers, enabling the robot to adapt to different objects with high accuracy while minimizing the time spent on image processing during navigation and cleaning operations.

Inventive Principle:
Principle #19Periodic action

4Productivity

If the cleaning robot descends the guard portion to push objects, then the productivity is improved by clearing obstacles, but the force required increases, potentially damaging the robot or objects

Engineering Contradiction:
Improveobstacle clearance capabilityVSAvoidpushing force on objects
Core Design Contradiction:
ProductivityVSForce

Solution Approach 1:

The guard portion is designed with local quality variations: it has a large surface area for distributed force application to prevent object damage, while the pushing action is localized to the front edge for effective obstacle clearance. The AI system also applies local quality by selecting push actions only for suitable objects (e.g., lightweight movable items) rather than all obstacles, improving productivity while controlling the force applied to each specific object type.

Inventive Principle:
Principle #3Local quality

Data Source

PatentEP3863813B1Cleaning robot and method of performing task thereof
Publication Date: 2024.03.20 SAMSUNG ELECTRONICS CO LTD
  • EP3863813B1 patent drawingFigure 1A~1B
  • EP3863813B1 patent drawingFigure 2~3A
  • EP3863813B1 patent drawingFigure 3B~4A

AI summary

A method, performed by a cleaning robot, of performing a task, is provided. The method includes capturing an image of an object in a vicinity of the cleaning robot, determining a task to be performed by the cleaning robot, by applying the captured image to at least one trained artificial intelligence (AI) model, controlling a guard portion to descend from in front of an opened portion to a floor surface of the cleaning robot, according to the determined task, and driving towards the object so the object is moved by the descended guard portion.