AI Moving Robot Region Segmentation for Efficient Cleaning Navigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing moving robots lack the ability to segment and recognize attributes of regions, leading to inefficient cleaning as they often repeat cleaning the same area or fail to return to their original location, due to their reliance on pre-stored maps and simple wall-following navigation.
Innovation Solution
A moving robot equipped with an image acquisition unit, sensors, and a controller that segments regions by acquiring and processing images to recognize attributes, allowing it to segment travel areas into distinct regions and recognize its current location, enabling efficient cleaning and navigation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the moving robot uses a pre-stored map generated by wall following to navigate, then the robot can avoid obstacles and perform cleaning, but the robot cannot accurately segment regions or recognize attributes of regions, leading to repeated cleaning of the same area
Solution Approach 1:
The patent segments the travel area into multiple distinct regions by dividing the map into a grid structure and identifying boundaries between regions. Each region is assigned a unique identifier, enabling the robot to recognize and differentiate between various areas, thus preventing repeated cleaning of the same location.
Solution Approach 2:
The patent replaces the traditional wall-following mechanical navigation method with an image-based recognition system. The robot uses an image acquisition unit to capture images of the environment, processes these images to identify region attributes, and uses this information for navigation, thereby achieving more precise region recognition compared to simple wall following.
2Device complexity
If the moving robot simply extracts an outline of the travel area using wall following, then the robot can generate a basic map, but the robot cannot segment multiple rooms into separate regions with recognized attributes
Solution Approach 1:
The patent divides the travel area map into a grid of cells and identifies region boundaries by analyzing transitions in the grid data. This segmentation approach allows the robot to distinguish between multiple rooms and areas, assigning each a unique region identifier while maintaining a manageable map structure.
Solution Approach 2:
The patent introduces an additional dimension of image processing to the traditional 2D map representation. By capturing images and processing them to extract region attributes, the system adds semantic information to the spatial map, enabling region recognition beyond simple geometric boundaries.
3Ease of operation
If the moving robot moves by converting direction when encountering obstacles without region segmentation, then the robot can continue cleaning, but the robot may fail to return to its original location or clean the same region repeatedly
Solution Approach 1:
The patent implements a feedback mechanism where the robot continuously monitors its current location by comparing it with the segmented region map. When the robot completes cleaning in a region or encounters an obstacle, it uses the region identification information to determine the appropriate next action, ensuring it can return to the original location and avoid redundant cleaning.
Solution Approach 2:
The patent replaces simple obstacle-avoidance direction conversion with a region-based navigation system. The robot uses image processing to identify region attributes and uses this information to make intelligent navigation decisions, thereby reliably returning to the original location and ensuring complete coverage without repetition.
Data Source
AI summary
Disclosed is an artificial intelligence moving robot including: a travel unit configured to move a main body based on a navigation map including a plurality of local maps; an image acquisition unit configured to acquire a plurality of images in regions corresponding to the plurality of local maps during movement; a storage unit configured to store the plurality of images acquired by the image acquisition unit; and a controller configured to recognize an attribute of a local map in which an N number of images photographed in multiple directions is acquired among the plurality of local maps, store the attribute recognition result in the storage, generate a semantic map composed of a predetermined number of neighboring local maps, and recognize a final attribute of a region corresponding to the semantic map based on attribute recognition results of the local maps included in the semantic map.