Irregular curved surface tracking method and system based on matrix feature learning neural network

Abstract

The invention discloses an irregular curved surface tracking method and system based on a matrix feature learning neural network. The method comprises the steps that an expectation feature vector is obtained; an initial matrix feature vector, the Jacobian matrix of the initial matrix feature vector and the target joint angular velocity vector of a mechanical arm are obtained; deep off-line training is conducted on a b-spline-based neural network controller by using the initial matrix feature vector, the Jacobian matrix of the initial matrix feature vector and the target joint angular velocityvector of the mechanical arm; feature errors of the current matrix feature vector and the expectation matrix feature vector are input to the trained b-spline-based neural network controller by the mechanical arm joint angular velocity vector to obtain the mechanical arm joint angular velocity vector under the current pose; the mechanical arm is controlled to move according to the mechanical arm joint angular velocity vector under the current pose to make a camera at the end of the mechanical arm to move along with the mechanical arm. By means of the method, the accurate positioning tracking ofan irregular curved surface can be achieved.

Description

technical field [0001] The invention belongs to the field of automatic control, and in particular relates to a tracking method and system for special-shaped curved surfaces based on moment feature learning neural network. Background technique [0002] With the rapid development of industry, the field of high-end manufacturing occupies a pivotal position in the national economy. The high-end manufacturing field includes aerospace, rail transit, new energy manufacturing, marine engineering equipment and other large-scale equipment manufacturing fields. Among them, a series of complex and special-shaped curved surface parts such as aero-engine turbine blades, ship propeller blades, and new energy turbine blades are among the most difficult parts to process in this field. Traditional manual processing of such parts often has disadvantages such as low processing precision and poor consistency of finished products, which lead to serious safety hazards in the use of parts. Using r...

AI Technical Summary

Problems solved by technology

[0010] The object of the present invention is to provide a method and system for tracking special-shaped curved surfaces based on moment feature learning neural network, which can realize positioning and tracking of special-shaped curved surfaces, overcome the poor robustness of hand-eye calibration accuracy in existing methods, and solve the problem of target depth in existing methods It is difficult to establish a function model and control problems caused by strong coupling of objects, nonlinearity, and serious time variation, to achieve precise positioning and tracking

Images