Robot-Mounted Vision Gripping for Precise Object Positioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for gripping objects in automated production lines face challenges such as inaccurate positioning due to varying object dimensions and orientations, requiring specific mechanical setups that are cumbersome to change when dealing with different object shapes or sizes, and are prone to contamination in backlit systems.
Innovation Solution
A method using a robot with a camera-mounted kinematics system that captures two-dimensional images of a search area with markings for perspective correction, allowing for real-time image processing and object location determination in a world coordinate system, enabling precise gripping without the need for elaborate calibration or stationary cameras.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a stationary camera with backlit search area is used for object detection, then object positioning can be achieved, but the system is highly susceptible to contamination from shaking objects and requires extensive calibration for different object types
Solution Approach 1:
The camera is merged with the robot's end effector, forming an integrated vision-gripping system. This eliminates the stationary camera setup and allows the camera to move with the robot arm, reducing contamination risk while maintaining positioning accuracy through real-time coordinate transformation between camera and robot reference frames.
Solution Approach 2:
The system transitions from a static camera setup to a dynamic mobile camera mounted on the robot. The camera's position and orientation change with robot movement, enabling flexible object detection without fixed contamination-prone lighting structures. Real-time coordinate transformations adapt to the dynamic camera position.
2Manufacturing precision
If mechanical alignment devices are used to prepare objects for gripping, then object positioning is achieved, but the mechanical equipment must be precisely adapted to each object type and requires extensive conversion measures when changing objects
Solution Approach 1:
Mechanical alignment devices are replaced with a vision-based system. The camera captures object images, and software algorithms determine object position and orientation. This substitution eliminates the need for mechanical adaptation when changing object types, as the vision system can detect and adapt to various geometries through image processing and coordinate transformation.
Solution Approach 2:
The system changes from fixed mechanical parameters to variable visual parameters. Instead of adjusting mechanical stops and alignment fixtures for each object type, the vision system captures images and computationally determines positioning parameters. Coordinate transformations between camera and robot reference frames enable flexible adaptation to different object dimensions and orientations.
3Ease of operation
If camera-based gripping with shaking objects is used to find suitable workpieces, then object finding capability is improved, but contamination susceptibility increases
Solution Approach 1:
The robot arm moves to predetermined approach positions and orientations before object detection. This preliminary positioning allows the camera to capture images from optimal angles without requiring violent shaking of objects. The systematic approach enables reliable object finding while minimizing contamination risk through controlled, gentle movements.
4Measurement precision
If two stationary adjustable cameras are used for 3D-based object recognition, then object positioning accuracy is improved, but system complexity and calibration requirements increase
Solution Approach 1:
The vision system extracts only the essential function of object detection from a complex multi-camera setup. A single camera mounted on the robot provides sufficient 3D positioning information when combined with coordinate transformation between camera and robot reference frames. This extraction simplifies the system while maintaining the core functionality of accurate object localization.
Data Source
Figure 1
Figure 2
Figure 3a~3d
AI summary
The present invention relates to a method for grasping objects in a search area using a robot calibrated in a world coordinate system, wherein the robot comprises a kinematic system, wherein a gripping device for grasping an object and a camera for capturing two-dimensional images of at least a sub-area of the search area are arranged on the kinematic system, wherein the search area has several markers that define a search area coordinate system and which are intended to determine perspective distortions in an image of the search area, wherein in a preliminary first calibration step, a first transformation function is determined based on a reference image of the search area, which converts the search area coordinate system into the world coordinate system. The method comprises aligning the camera with the search area, capturing a two-dimensional image of the search area,a rectification of the image based on markers captured in the image, a determination of the position and orientation of an image of an identified object present in the rectified image in an image coordinate system, a determination of the position and orientation of the object in the search area coordinate system based on the position and orientation of the image of the object in the image coordinate system and based on images of the markers, a determination of the position and orientation of the object in the world coordinate system based on the position and orientation of the object in the search area coordinate system and based on the first transformation function, and a control of the kinematics to move the gripping device into a gripping position in which the identified object is gripped by the gripping device, based on the position and orientation of the object in the world coordinate system, wherein at least one of the steps,which include capturing the image, correcting the image, and determining the position and orientation in the image, search area, and/or world coordinate system, performed during a movement of the kinematics. The invention further relates to a control unit and a positioning system.