Robot Pose Localization Using Image-Laser Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing robot locating methods, particularly in autonomous cleaning devices, suffer from inaccurate positioning and fail to detect hijacking events, leading to reduced efficiency and continued operation in incorrect locations.
Innovation Solution
Employ a laser distance sensor using a triangular ranging method combined with a camera for image data, particle filtering, and SLAM algorithms to enhance location determination, incorporating image and ranging data to improve accuracy and detect hijacking events.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional SLAM algorithms are used for robot positioning, then the robot can perform autonomous navigation, but the positioning accuracy is insufficient and hijacking events cannot be detected
Solution Approach 1:
The patent combines multiple positioning methods (visual positioning using image collection units, laser positioning using distance measurement units, and inertial positioning using sensors) into a unified positioning system. This multi-source fusion approach resolves the contradiction by simultaneously improving positioning accuracy through multiple data sources and enabling hijacking detection through cross-validation of positioning results from different methods
Solution Approach 2:
The patent implements a feedback mechanism where the processor continuously compares positioning results from different methods and provides feedback to adjust positioning accuracy. When discrepancies are detected between visual, laser, and inertial positioning results, the system triggers re-positioning or hijacking event detection, thereby resolving the contradiction between accuracy and reliability
2Measurement precision
If multiple sensors and positioning methods are integrated, then positioning accuracy improves, but system complexity increases
Solution Approach 1:
The patent segments the positioning system into distinct functional modules: image collection units for visual positioning, distance measurement units for laser positioning, and sensors for inertial positioning. Each module operates independently and can be selectively activated based on environmental conditions, thereby improving positioning accuracy while managing system complexity through modular architecture
Solution Approach 2:
The patent implements dynamic selection of positioning methods based on environmental conditions and robot state. The processor dynamically adjusts which positioning methods are active and how they are weighted in the fusion algorithm, allowing the system to achieve high accuracy when needed while reducing complexity by deactivating unnecessary sensors in certain situations
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enhances locating accuracy and efficiency by accurately determining the robot's position and detecting hijacking events, ensuring optimal cleaning operations.
Implementation Method 1
a laser distance sensor using a triangular ranging method
Data Source
Figure 1~2
Figure 3A
Figure 3B
AI summary
A positioning method and apparatus (2200) for a robot (100), and a computer-readable storage medium. The positioning method comprises: determining current possible pose information of a robot (100) according to current distance measurement data acquired by a distance measurement unit (202); determining, according to current image data acquired by an image acquisition unit, historical image data matching the current image data (204), the historical image data being acquired by the image acquisition unit at a historical moment; obtaining historical pose information of the robot (100) at the acquisition moment of the historical image data (206); and when the number of pieces of the current possible pose information is two or more, matching the historical pose information with each piece of the current possible pose information, and using the matched current possible pose information as current target pose information (208).