Cleaning Robot Image Processing for Trash and Spillage Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional cleaning robots are unable to distinguish between trash and spillage, requiring manual user intervention to set appropriate cleaning modes.
Innovation Solution
A system comprising an image capturing module, cleaning devices, and a processor that uses anomaly detection, object detection, and segmentation models to identify and differentiate between regular and irregular targets, controlling cleaning devices accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If conventional cleaning robots use manually set working modes, then the cleaning robot can perform basic cleaning tasks, but the cleaning robot is unable to identify and distinguish between various types of targets such as trash and spillage
Solution Approach 1:
The patent replaces manual mode selection with automated image processing and AI-based target recognition systems. The cleaning robot uses cameras and processors to automatically identify and classify targets (trash, spillage) and autonomously selects appropriate cleaning modes, eliminating the need for manual intervention while enhancing adaptability to different target types.
2Reliability
If the cleaning robot requires manual user intervention to set cleaning modes, then the cleaning robot can clean targets appropriately, but the user must manually identify and classify different target types
Solution Approach 1:
The cleaning robot performs self-identification and self-classification of targets using integrated image capturing modules and AI processing systems. The robot autonomously determines the type of target (trash, spillage, liquid) and automatically selects the appropriate cleaning mode without requiring user intervention, making the system both reliable and easy to operate.
3Measurement precision
If the cleaning robot processes images with high precision to distinguish target types, then the cleaning robot can accurately identify trash and spillage, but the processing time and computational resources increase
Solution Approach 1:
The patent implements a multi-stage image processing pipeline that segments the analysis into distinct phases: initial anomaly detection to identify potential targets, followed by shape analysis to classify targets as regular or irregular, and finally detailed attribute recognition. This segmented approach reduces computational complexity at each stage while maintaining high identification accuracy, thereby reducing overall processing time.
Data Source
AI summary
According to various embodiments, there is a system comprising: an image capturing module configured to capture an image of an area, the image comprising a target to be disposed of; at least one cleaning device configured to perform a task to dispose of the target; and a processor configured to: receive the image from the image capturing module; determine an attribute of the target and a location of the target from the image; and control the at least one cleaning device to perform the task, based on the determined attribute of the target and the determined location of the target, wherein the processor is further configured to detect a rough shape of the target from the image, and process the image in a different manner based on whether the target is of regular shape or irregular shape to determine the attribute of the target and the location of the target.