Homography matrix based visual servo control method for shortest path

Abstract

The invention discloses a homography matrix based visual servo control method for a shortest path. The method comprises steps as follows: motion of a robot is decomposed into three stages, firstly, the robot is rotated to face a position point where a target image is located, the position point where the target image is located is called a target point for short, then the robot is horizontally moved to the target point, finally, the robot is rotated to an angle of a target pose, and the robot motion can be controlled according to corresponding control laws at each stage. According to the method, elements in a homography matrix are directly utilized to constitute error functions, homography matrix decomposition is not required, robustness for camera parameters and image noise is better, and meanwhile, the problems about unknown depth information of a monocular camera and nonholonomic constraint of a tracked robot are solved, so that the robot reaches the target pose stably through the shortest path.

Description

technical field [0001] The invention relates to a visual servo control method, in particular to a shortest path visual servo control method based on a homography matrix. Background technique [0002] Automatic object-seeking robots have been gradually used in industrial production. Existing automatic object-seeking robots usually use high-precision positioning of known maps to find objects, which has poor flexibility and high implementation costs. Some robots adopt the servo control method based on visual information to control the robot motion to improve flexibility and stability. The feedback information of the image-based visual servo control system is defined on the image plane, and its control goal is to minimize the error between the currently acquired image state s and the given target image state s*. Using the relevant knowledge of machine vision, the corresponding feature points can be obtained by matching the current image and the target image, and the pixel coord...

AI Technical Summary

Problems solved by technology

The early research on nonholonomic constraints mainly used differential equation theory and variational principles to model and analyze systems, etc. Later, some scholars applied mathematical tools such as differential geometry and group theory to nonlinear system analysis, making nonholonomic Systematic research has developed rapidly, but the process of analysis is still relatively complicated

[0004] Existing automatic object-finding robots usually use high-precision positioning of known maps to find objects, which has poor flexibility and high implementation costs. Some robots controlled by visual servoing require binocular cameras to calculate depth information, which makes calculations complicated and high implementation costs

Images