Cleaning robot and task performing method therefor
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing cleaning robots are limited in their ability to specifically identify and respond to objects around them due to a limited combination of sensors, often relying on repetitive avoidance patterns without detailed information about target objects.
Innovation Solution
A cleaning robot equipped with a plurality of sensors, a camera, and a processor that photographs objects, applies images to a trained artificial intelligence model for recognition, and uses sensor data to obtain additional information about objects, allowing for tailored tasks based on object recognition and additional sensor data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a limited sensor combination is used, then the device complexity is reduced, but the object recognition precision deteriorates
Solution Approach 1:
The patent segments the sensing system into multiple independent sensor modules, each responsible for detecting specific object attributes. The sensor controller selectively activates different sensors based on the cleaning task requirements, allowing precise object recognition while managing system complexity through modular organization.
Solution Approach 2:
The patent implements a universal sensing architecture where a single sensor controller manages multiple sensor types (contact, non-contact, proximity sensors). This multi-functional controller can adaptively select and coordinate different sensors based on the cleaning task, eliminating the need for separate control systems for each sensor type.
2Measurement precision
If multiple sensors are used to obtain additional information, then the object recognition precision is improved, but the device complexity increases
Solution Approach 1:
The patent employs dynamic sensor selection where the sensor controller adaptively activates specific sensors based on the cleaning task and object characteristics. Instead of all sensors operating continuously, the system dynamically configures the active sensor set, reducing complexity while maintaining high recognition precision when needed.
Solution Approach 2:
The sensor controller acts as an intermediary that coordinates multiple sensors and integrates their data. This intermediary component manages the complexity by providing a unified interface between diverse sensors and the task execution system, allowing precise object recognition through coordinated multi-sensor operation.
3Adaptability or versatility
If sensor data processing is performed to obtain additional information, then the task suitability is improved, but the processing time increases
Solution Approach 1:
The patent performs preliminary data processing by the sensor controller to extract essential object characteristics before task execution. By pre-processing sensor data to identify key features (object type, size, position), the system reduces the time required for real-time task adaptation while maintaining high task suitability.
Solution Approach 2:
The system implements feedback loops where the sensor controller continuously monitors object information and adjusts task execution in real-time. This feedback mechanism allows the robot to adapt tasks based on processed sensor data without significant time loss, as the processing is integrated into the continuous operation cycle.
Data Source
AI summary
A task performing method for a cleaning robot, according to the disclosure, comprises the operations of: photographing an object in proximity to the cleaning robot includes obtaining recognition information of the object included in the photographed image, by applying the photographed image to a trained artificial intelligence model, obtaining, from among a plurality of sensors, additional information of the object by using a result obtained by detecting the object by at least one sensor selected, based on the recognition information of the object, and performing a task of the cleaning robot based on the additional information of the object. The trained artificial intelligence model, for example, a deep learning neural network model, in which a plurality of network nodes having weights are located in different layers so as to exchange data according to a convolution relationship, can be used, but is not limited to the aforementioned example.