Optimal contact line detection method between robot parallel plate clamp and target object

Abstract

The invention discloses an optimal contact line detection method between a robot parallel plate clamp and a target object, and relates to the field of intelligent robot vision. The method comprises the steps of (1) obtaining a depth image of a target object; (2) carrying out convolution filtering on the depth image; (3) carrying out two-dimensional convolution to calculate the object gradient; (4) calculating the amplitude of the gradient and the direction of the gradient; (5) arranging and counting the gradient directions by using a histogram, and selecting a main direction angle; (6) tracking the contour line of the gradient image so as to determine the length of the contour line; (7) removing the contour line exceeding a specified threshold value; (8) checking the grabbing stability of the contact line pair; (9) selecting a contact line pair with the highest reliability score as an optimal grabbing position; and (10) grabbing positions of which the reliability is greater than a threshold value, and selecting the highest score for a plurality of grabbing positions. According to the invention, the problem of accurate and efficient detection of the optimal contact line between the robot parallel plate clamp and the target object is solved.

Description

technical field [0001] The invention relates to the field of intelligent robot vision technology, in particular to a method for detecting an optimal contact line between a parallel plate fixture and a target object based on a depth image. Background technique [0002] Robot vision intelligent grasping is currently a hot research direction in the field of robotics. The application of machine vision in industry is becoming more and more extensive with the continuous improvement of industrial automation and intelligence. [0003] At present, many robots use RGB cameras to obtain scene and target object data, and then use image processing to extract the grasping position of the target object. [0004] The current method of extracting the grasping position of the target object is mainly to obtain feature points through image processing technology, such as SIFT, SURF, PFH, FPFH, Harris, etc. [0005] If we use the current method to extract the grasping position of the object, we ...

AI Technical Summary

Problems solved by technology

[0005] If we use the current method to extract the grasping position of the object, we have to traverse various directions to select and detect features, which will affect the detection efficiency of the contact position of the target object. Therefore, it is necessary to accurately extract and speed up the detection of the grasping position. important subject

Images