Mechanical arm target tracking method based on visual servo

Abstract

The embodiment of the invention discloses a mechanical arm target tracking method based on visual servo. The method comprises the following steps: an image acquisition and feature recognition module acquires image features of a target in real time by using an industrial camera within an image recognition cycle, and detects image feature coordinate information; a visual servo operation module carries out image Jacobian matrix online estimation according to the image feature coordinate information, and the space velocity of the tail end of a mechanical arm at the current moment is calculated; a mechanical arm motion controller module converts the space velocity of the tail end of the mechanical arm at the current moment into a joint target angular velocity within a mechanical arm control cycle, and mechanical arm control cycle time constant integral is used as a mechanical arm joint target angle; and when the mechanical arm control cycle time constant integral is used as the mechanical arm joint target angle, a speed feed-forward compensation module dynamically compensates the mechanical arm joint target angle, the mechanical arm joint target angle is adjusted in real time, and target tracking is completed.

Description

technical field [0001] The present invention relates to the field of machine vision, and more specifically, to a method for tracking a manipulator target based on visual servoing. Background technique [0002] The field of industrial robots has put forward new requirements for the degree of intelligence and flexibility of the manufacturing industry. The market demand for industrial robots has gradually changed from traditional industrial robot structured scenarios, repetitive operations, and application scenarios with poor flexibility. Style, small batch, The replacement of intelligent production methods puts forward new requirements for industrial robots to have the ability to perceive changes in the external environment, and vision-based servo control technology can solve key problems in some scenarios, such as end-assisted flexible assembly and space target docking. [0003] Visual servo control technically includes image-based, position-based, and 2.5D-based visual servo...

AI Technical Summary

Problems solved by technology

However, in this method, the SVM algorithm consumes a lot of time, and the visual recognition cycle will increase, resulting in a much longer than the robot control cycle. The system control does not meet the real-time requirements, and the dynamic lag of the servo system will increase.

Moreover, the results obtained by SVM training are very dependent on the selection of training samples, and the calculation accuracy of the Jacobian matrix will be affected by the samples and produce deviations.

Images