Monocular Camera Position Control for Connector Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
It is challenging to control the position for assembly using only information from a monocular camera, as existing methods require complex image processing and specific design for each connector type, limiting their versatility and efficiency.
Innovation Solution
A position control device that employs a monocular camera and processing circuitry with multiple neural networks, where the imaging circuitry captures images of two objects, and the processing circuitry selects the appropriate neural network based on the control amounts outputted, allowing the drive unit to move the first object to align with the second object accurately.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complex image processing algorithms are developed for each connector type, then position control accuracy is improved, but device complexity and design cost increase significantly
Solution Approach 1:
The patent applies universality by using a single monocular camera system that can handle multiple connector types without requiring specific image processing algorithms for each type. The system achieves this by capturing images at multiple focal positions and using focus analysis to determine 3D information, making the system versatile across different connector geometries while maintaining position control accuracy.
Solution Approach 2:
The patent transitions from 2D image analysis to 3D position determination by capturing images at multiple focal positions (different depths). By analyzing which focal position produces the sharpest image, the system infers depth information and calculates 3D coordinates, thereby achieving accurate position control without complex algorithms for each connector type.
2Measurement precision
If specific design and algorithms are created for each connector type, then alignment precision is improved, but adaptability and versatility decrease
Solution Approach 1:
The system achieves universality by using a monocular camera with multi-focal plane imaging capability that works with various connector types without requiring type-specific algorithms. The focus-based depth estimation method is applicable to any connector geometry, maintaining alignment precision while enabling broad adaptability across different connector designs.
Solution Approach 2:
The system employs self-service by automatically determining 3D position information through focus analysis of captured images. The system itself identifies the focal position that produces the sharpest image and uses this information to calculate depth, eliminating the need for external calibration or type-specific configuration for different connectors.
3Measurement precision
If multiple devices such as force sensor and stereo camera are used, then position errors are reduced, but device complexity and cost increase
Solution Approach 1:
The patent makes the monocular camera multi-functional by enabling it to perform both 2D image capture and 3D depth estimation through multi-focal plane imaging. By capturing images at different focal positions and analyzing focus sharpness, the single camera device replaces the functionality of multiple devices (stereo camera or force sensor + camera), reducing system complexity while maintaining position accuracy.
Solution Approach 2:
The patent introduces focus sharpness analysis as an intermediary method to extract depth information from 2D images. By using the focus position that produces the sharpest image as a mediator, the system indirectly obtains 3D position data without requiring direct 3D sensing devices, thereby simplifying the hardware configuration while achieving accurate position control.
Data Source
AI summary
An imaging unit, a control parameter generation unit, a control unit, and a drive unit are provided. The imaging unit captures an image including two objects. The control parameter generation unit feeds information of the captured image including the two objects into an input layer of a neural network, and outputs a position control amount for controlling the positional relation of between the two objects as an output layer of the neural network. The control unit controls current or voltage to control the positional relation between the two objects by using the outputted position control amount. The drive unit changes a position of one of the two objects by using the current or the voltage. Here, the control parameter generation unit selects the neural network from a plurality of neural networks. Therefore, even if there are differences between objects or errors in the positional relationship between the two objects, alignment can be performed more accurately.