Mechanical arm grabbing control method based on binocular vision

Abstract

The invention relates to the technical field of kinova mechanical arms and ZED vision, and provides a mechanical arm grabbing control method based on binocular vision. The method comprises following steps: (1), constructing an experimental platform; (2), recognizing the positions and the postures of target objects when the target objects are placed obliquely; (3), recognizing the positions and thepostures of the target objects when the target objects are placed horizontally; (4), recognizing and grabbing the target objects by a mechanical arm end executor; and (5), reaching a designated jointangle state after mechanical arms grab the targets. The method comprises following advantages: firstly, the posture angles of the target objects are estimated, included angle deviation between the direction vector and the coordinate axis plane is in a range of -5 degrees to 5 degrees, and accuracy of detection is greatly improved; secondly, with adoption of color and shape design, disturbance terms with the same or close color can be prevented from disturbing the target object recognition; and thirdly, double mechanical arms are arranged to grab the target objects on the left and right, and finally, the effect that the left mechanical arm and the right mechanical arm grab the postures of the target objects is realized.

Description

technical field [0001] The invention relates to a binocular vision-based grasping control method for a robotic arm, which belongs to the technical field of kinova robotic arm and ZED vision. Background technique [0002] Binocular vision pose measurement is an important target pose method, but it is difficult to measure the pose of an obliquely placed object. Some existing methods include: circle extraction method and feature point extraction method. [0003] Circular section extraction method: the existing method is to use three industrial CCD cameras to extract the center of the upper and lower sections of the cylindrical target, define a coordinate system, and calculate the angle between the line connecting the two section centers and the coordinate system to describe the position Attitude, the accuracy of this method is greatly affected by the shape of the target, and the center of the occluded circular section is not easy to extract. [0004] Feature point extraction m...

AI Technical Summary

Problems solved by technology

In order to solve the problems of difficulty in extracting the cross-section circle and low precision, the present invention proposes the method of using foreground extraction to obtain the apex of the cross-section arc of the target object, and estimate the posture of the target object with the coordinate difference of the apex, and finally control the manipulator to be as flexible as a human hand. grab target

Images