Mechanical arm real-time tracking method based on binocular vision guidance

Abstract

The invention discloses a mechanical arm real-time tracking method based on binocular vision guidance, belongs to the technical field of robot vision, and relates to the field of industrial robots. The method comprises the steps: carrying out binocular stereo vision calibration, correction and matching and mechanical arm forward and reverse kinematics modeling; secondly, establishing a coordinateconversion relationship between the binocular camera and the mechanical arm, and quickly acquiring an image sequence through the binocular camera; processing an image sequence acquired by the camera by utilizing linear kernel correlation filtering, converting the image sequence into a robot base coordinate system {B}, and obtaining mechanical arm motion control parameters through kinematics calculation; finally, enabling the mechanical arm controller, the mechanical arm body and the actuator to complete related operation tasks. The tracking method provided by the invention can adapt to the change of the workpiece scale, considers the instantaneity and robustness, greatly weakens the influence of uncertainty factors on the system, and has the characteristics of high intelligence and high systematization degree.

Description

technical field [0001] The invention belongs to the technical field of robot vision, and relates to a real-time tracking method of a manipulator based on binocular vision guidance in the field of industrial robots. Background technique [0002] At present, the intelligent manufacturing model represented by robot technology is developing rapidly. Machine vision perception is a high-precision, non-contact and widely used technology, which can greatly improve the automation and intelligence of production. However, the traditional manipulator uses trial teaching or offline programming to control the manipulator to complete processing, assembly, handling, grasping and other tasks in accordance with the prescribed actions. In a complex and changeable environment, the position and posture of the workpiece cannot be determined. It is impossible to locate in real time. Provide the position and orientation of the workpiece to the robotic arm according to the visual information and gu...

AI Technical Summary

Problems solved by technology

"Visual manipulator control device and method based on Camshift visual tracking and D-H modeling algorithm" disclosed by Han Jiuqiang of Xi'an Jiaotong University, patent number 201510394465.4. Although the traditional Camshift algorithm has low algorithm complexity, it is suitable for complex backgrounds or rich textures. Object tracking is poor

Generative algorithms ignore prior background information. Although discriminative algorithms significantly improve tracking performance, they are mainly used in video target tracking, and are rarely used in actual industries. They cannot provide solutions for robotic arm vision guidance.

Images