Mobile Manipulator Control Based On Machine Vision

In the production life, the multi-degree of freedom mechanical arm with its flexible, up to large range, high load capacity and other characteristics, plays a very significant role in a wide range of applications. The rapid development of "Vision-based robot control technology, making manipulator working independently in an unknown environment become a possible. With the rapid development of industrial automation as well as the high-tech industry, the amount of mechanical arm continues to increase, and its application areas are also expanding, largely promote the development of Vision-based multi-degree of freedom manipulator control technology.This article, relying on wireless Wi-Fi based embedded wheeled robot research platform, researches and practices the mobile manipulator system to machine vision orientation and target clip problems. Mainly solves the problem of how to use the space position feedback information obtained by binocular machine vision to control the mechanical arm automatically reaching target locations in space, clipping the target in the end.This paper firstly completed the overall design scheme of machine vision mobile manipulator system, separately designed the work from two aspects of hardware and software platform of the system. The introduction of hardware platform mainly includes mobile robot chassis, embedded development board, mechanical arm, motor and mechanical arm controllers, hardware parameters of binocular machine vision module and other modules, as well as the specific reasons for the selection of each module. For the introduction of software platform, briefly introducing of the embedded operating system used by the server in this mobile manipulator control system; completing a binocular machine vision module driver writing; and finally developing the operation interface of this system. Hardware and software platform together constitute the functionality of the system and lay a solid foundation for the mobile manipulator system to achieve the target accurately clipping.Secondly, this article studied the principles of the binocular machine vision module obtaining the object space coordinate. Setting up an optical model of parallel binocular vision. This model shows that getting the spatial three-dimensional information of the object needs to know the internal parameters of binocular machine vision module, and a single camera coordinate information provided by the object in an image; On the basis of it, firstly using toolbox-calib to calibrate camera, and obtaining the internal parameters of binocular machine vision module acquired by the optical model; And then presenting a target detection method based on Hough transform, obtaining the plane position information which two cameras capture a single image of the object, and lastly getting the space position information of the target.Thirdly, solving the problem of how to control the end effector to reach the target location under the condition of knowing space position information of target. Introducing the method of describing spatial position and a universal modeling method of mechanical arm kinematics model ---D-H method; then analyzing the forward kinematics of the model, using MATLAB software to find out the working range of the model; lastly, giving a method to calculate the Inverse problem of mechanical arm kinematics. At the same time, using the method of simulation and experiment to verify the reliability of the mathematical model, providing a basis for the robot platform moving to the target, the target location, and final clipping.Finally, from the process of operation system, introducing and analyzing the mobile manipulator control system based on machine vision and the operation and the use of the system is introduced, using this system to do the target clipping experiment, through the analysis of experimental results and the experimental error, illustrating the reliability, high accuracy and high efficiency of the system.