Mobile Robot Map Creation Using Visual Calibration Key Frames
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing cleaning robots relying on inertial navigation systems face inaccuracies due to component errors and environmental complexities, leading to poor positioning and map creation, which affects path planning.
Innovation Solution
A map creation method for mobile robots using an odometer, gyroscope, and camera, employing the ORB algorithm for image matching to calibrate coordinates and reduce errors, combined with a path planning method that calibrates odometer and gyroscope data at key points to improve navigation accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If inertial navigation system (odometer and gyroscope) is used for positioning and map creation, then the mobile robot can navigate autonomously, but accumulated errors increase with working time leading to inaccurate positioning and map information
Solution Approach 1:
The patent implements a feedback mechanism by capturing images at key frames, extracting feature points, and comparing them with pre-stored reference images. The coordinate transformation between current and reference coordinate systems provides continuous feedback to correct accumulated inertial navigation errors, maintaining positioning accuracy over time.
Solution Approach 2:
The patent replaces pure mechanical inertial navigation (odometer and gyroscope) with a hybrid system that incorporates visual feedback from cameras. By substituting mechanical measurement with optical field-based feature matching and coordinate transformation, the system eliminates accumulated errors without relying solely on mechanical components.
2Measurement precision
If feature point extraction and matching is performed at every position, then positioning accuracy is improved, but computational complexity and time consumption increase significantly
Solution Approach 1:
The patent applies preliminary action by pre-storing reference images and their corresponding coordinate information before the robot begins navigation. This pre-prepared reference data enables rapid feature matching during operation, avoiding the need for real-time image processing at every position and significantly reducing computational time.
Solution Approach 2:
The patent uses partial action by performing feature point extraction and matching only at key frames rather than at every single position. This selective approach maintains positioning accuracy while dramatically reducing the computational burden and time consumption compared to continuous full-area processing.
3Area of stationary object
If the mobile robot moves in complex patterns to cover the working area, then map coverage is improved, but the friction coefficient fluctuation and sliding errors increase affecting inertial navigation accuracy
Solution Approach 1:
The patent introduces visual feature matching as an intermediary mechanism between the robot's movement and positioning system. The feature point extraction and coordinate transformation act as a mediator that decouples positioning accuracy from movement stability, allowing the robot to navigate complex patterns without transmitting sliding and friction errors directly to the positioning system.
Data Source
AI summary
The present disclosure discloses a map creation method of a mobile robot, including establishing a rectangular coordinate system in a working area of the mobile robot; causing the mobile robot moving in a character shape in the working area; obtaining a key frame picture of the mobile robot at a point Pi and saving the picture and coordinates of the point Pi; obtaining a picture of the mobile robot at a point P′i, wherein an abscissa or ordinate of the point P′i is the same as that of the point Pi; extracting and matching features of the pictures acquired at the point Pi and the point P′i; calibrating coordinates of the mobile robot at the point P′i and data of the odometer and/or the gyroscope according to the matching result and saving them; and repeating the obtaining to the calibrating until the map creation in the working area is complete.