Rapid and high-accuracy NAO type soccer robot visual processing method

Abstract

The invention discloses a rapid and high-accuracy NAO type soccer robot visual processing method which comprises the following specific implementation steps: firstly, operating a BHuman robot drive, synchronously operating a software system at the same time, and carrying out image information transmission with a robot; then, calculating and processing images by a well packaged modular algorithm in a driving core algorithm processing library; for a test mode, firstly starting a system daemon process, calling a GUI graphical debugging interface to display an image and output information, and driving a well packaged core algorithm processing module to calculate and process the images. Computer vision and instruction set optimization technology is utilized to carry out rapid assembly line processing and operation on image signals, so that a high-performance robot vision processing algorithm process is realized, and a software system realized by the vision processing method can be simply and rapidly applied to an NAO robot and has high practicability. And the accuracy and the efficiency of robot football match visual processing are greatly improved.

Description

technical field [0001] The invention relates to the technical field of robot vision processing, in particular to a fast and high-accuracy NAO type soccer robot vision processing method. Background technique [0002] In most of the current visual processing workflows, existing computer vision libraries are widely used, and these existing libraries often have defects such as too long codes, insufficient performance, and inability to meet customization requirements. At the same time, existing algorithms are often susceptible to interference caused by light changes, which causes errors to accumulate continuously in the cascade processing process, and the final results are far inferior to expected results. For the field of object classification, most of the existing technologies are based on image space High-order gradient features for detection and analysis are not very robust. [0003] In the football robot RoboCup standard platform group competition, the robot's real-time per...

AI Technical Summary

Problems solved by technology

However, the traditional scheme's high dependence on color and lighting makes it not have good usable effects in most scenarios.

At the same time, the decentralized and highly coupled processing algorithm structure is not conducive to the modularization and subsequent further expansion of the overall project

At present, most of the traditional methods are lengthy and have insufficient performance. It is difficult to meet the extremely high real-time requirements of football robot games. In addition, the traditional methods have low efficiency in division of labor and cooperation. Traditional algorithms make it more difficult for algorithm researchers to write and debug algorithms. Algorithm writers need to deal with a lot Memory, technology, and framework issues that have nothing to do with the algorithm itself,

Images