An image processing method, platform and system based on cloud robot vision

Abstract

The invention discloses an image processing method, platform and system based on cloud robot vision. The method includes a host in a distributed server cluster constructed by a public cloud server to detect in real time the storage file information of each slave; the host receives in real time the information sent by the robot The encoded video frame image; according to the storage file information of the slave, the host assigns the corresponding file storage task to each slave, and stores the encoded video frame image in the corresponding slave; the host assigns the image to each slave Recognition tasks, wherein, the slave machine is equipped with an image recognition engine that encapsulates several image recognition algorithms; the slave machine receives the image recognition task and invokes the corresponding image recognition algorithm in the image recognition engine to perform the image recognition task, and the recognition result is sent to the host machine Feedback to the robot.

Description

technical field [0001] The invention belongs to the field of robots, and in particular relates to an image processing method, platform and system based on cloud robot vision. Background technique [0002] With the development of science and technology, robot vision technology has been greatly developed, and the corresponding visual recognition algorithm has gradually matured. However, problems such as low sampling rate, large amount of image data, high algorithm complexity, and long running cycle of the robot vision system hinder the popularization and application of robot vision. With the continuous maturity of cloud technology, the concept of cloud robot has emerged. Cloud robot is the combination of cloud computing and robotics. Just like other network terminals, the robot itself does not need to store all data information or have super computing power, but can connect to relevant servers and obtain the required information when needed. This method is an important way t...

AI Technical Summary

Problems solved by technology

[0003] (1) The traditional robot body carries a large amount of computing and storage equipment, which leads to the contradiction between the function of the robot and the lightweight and low cost of the robot itself;

[0004] (2) The traditional robot visual recognition platform adopts the stand-alone training mode of high-performance computers, which leads to the inability to reuse the system code and it is difficult to solve the code usage problem of heterogeneous robots;

[0005] (3) Traditional parallel computing mostly relies on high-performance computer clusters. High-performance computers are replaced too quickly, resulting in high costs and a relatively fixed number of server nodes, which is not flexible enough. Traditional private image recognition platforms lack relevant network security protection, which is very easy Cause hackers, viruses and other interference, and platform server hardware maintenance and management, waste a lot of manpower and material resources

[0006] The above deficiencies lead to problems such as low image recognition efficiency, long cycle, and inconsistent data structure and protocol of robots, which hinder the application and promotion of vision in the direction of robots.

Images