Monocular Mobile Robot Pose Control Using Spherical Image Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing mobile robot control methods rely on expensive and bulky sensors like lidar, RGB-D, and stereo cameras for pose measurements, which are costly and complex, and require prior knowledge of the target model for monocular camera-based pose reconstruction.
Innovation Solution
An image-based control method using a processor and camera to extract invariant image features and rotation vector features from matching feature points projected onto a virtual unitary sphere, allowing the robot to adjust its pose without requiring pose measurements or prior knowledge of the target model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If expensive and bulky sensors like lidar, RGB-D, and stereo cameras are used for pose measurements, then measurement precision is improved, but device complexity and manufacturing cost increase
Solution Approach 1:
The patent extracts the essential function of pose measurement from complex sensor systems and implements it through a simplified monocular camera system combined with feature point extraction and projection methods. Instead of using multiple sensors (lidar, RGB-D, stereo cameras), the solution extracts only the necessary visual information from a single camera to achieve pose estimation through image feature matching and geometric projection.
Solution Approach 2:
The patent replaces expensive, complex sensor systems with a simple, inexpensive monocular camera. The solution uses basic image processing and geometric calculations rather than costly hardware, achieving comparable functionality with significantly reduced manufacturing cost and system complexity.
2Device complexity
If monocular camera-based pose reconstruction is used, then device complexity is reduced, but computation complexity increases due to requirements for prior knowledge of target model and complex pose estimation
Solution Approach 1:
The patent extracts only the essential visual features from images (feature points) and their projections, eliminating the need for complex pose estimation algorithms and prior knowledge of target models. The solution focuses on extracting invariant geometric relationships from images that can directly guide robot motion without requiring full pose reconstruction.
Solution Approach 2:
The patent transforms the problem from estimating complex pose parameters (position and orientation in 3D space) to measuring simpler projection parameters (2D image coordinates and their relationships). By changing the parameters from full pose estimation to projection-based feature point relationships, the computational complexity is significantly reduced while maintaining the ability to control robot motion.
3Manufacturing precision
If Cartesian coordinate system-based control is used, then control precision is improved, but adaptability decreases due to dependency on inertial reference systems
Solution Approach 1:
The patent inverts the traditional control approach by instead of controlling the robot to reach a specific pose in inertial space, it controls the robot to achieve specific projection relationships of feature points in the image plane. This inversion from pose-based control to projection-based control eliminates dependency on inertial reference systems while maintaining control precision through invariant geometric relationships.
Solution Approach 2:
The patent creates a universal control method that works with any visual scene containing detectable feature points, without requiring specific knowledge of the target object or its relationship to inertial frames. The projection-based approach is universally applicable to different environments and objects, enhancing adaptability while maintaining precision through geometric invariance.
Data Source
AI summary
A mobile robot control method includes: acquiring a first image that is captured by a camera on a robot when the robot is in a desired pose; acquiring a second image that is captured by the camera on the robot when the robot is in a current pose; extracting multiple pairs of matching feature points from the first image and the second image, and projecting the extracted feature points onto a virtual unitary sphere to obtain multiple projection feature points, wherein a center of the virtual unitary sphere is coincident with an optical center of coordinates of the camera; acquiring an invariant image feature and a rotation vector feature based on the multiple projection feature points, and controlling the robot to move until the robot is in the desired pose according to the invariant image feature and the rotation vector feature.