DESIGN OF INTELLIGENT ROBOTIC CELL WITH CAMERA SYSTEM

Urgency of the research. More complex robotic systems are characterized by a certain degree of intelligent behavior where, based on input, the system is able to adapt its behavior. The implementation of elements that support intelligent behavior in robotic systems, especially those based on the image of devices, is becoming common practice. The reason is simple, such a system is faster and more accurate. Target setting. Creating machine vision, however, is a complex problem, especially when it comes to applications with non-standard requirements. For each task, the vision system needs to be adapted to the conditions and requirements of the monitored objects. Other image adjustments and algorithms need to be applied to static objects rather than moving objects.Two-dimensional image information is sufficient for some manufacturing process, while others require a third dimension to remove a given piece from a disordered pile. Creating an intelligent robotic cell with a camera system therefore requires the creation of a vision system that meets the specified requirements. This is where space is open, because there are many different procedures and principles to deal with, but not all are equally effective and reliable. Actual scientific researches and issues analysis. Many of the image processing methods can be combined with each other, or a new, better way to solve the problem can be developed using the approaches already known. Adding to this fact non-standard requirements profiled in practice, there is an undeniable reason why it is appropriate to deal with image processing for industrial use. Uninvestigated parts of general matters defining are designing and create a robotic cell, whose activity will be controlled on the basis of image perception obtained by digital camera.The research objective of this article is to design and create a robotic cell, whose activity will be controlled on the basis of image perception obtained by digital camera. The obtained image will be subjected to suitable image processing algorithms which will result in the generation of control instructions for controlling the manipulator movement.The statement of basic materials. The work deals with the design of a robotic cell whose task is to manipulate sample objects placed on the conveyor belt by means of a parallel manipulation robot based on image perception. The main part of the design is the creation of control software, which in the first level ensures the proper functioning of the individual components and in the second level their mutual cooperation, which ensures the performance of the required functionality of the robotic system as a whole. Created software runs on Windows 7 operating system, where it offers a simple tool to control the movement of the arms of a parallel robot without using other control means. This means that the robot's movements can be controlled directly from the control program, allowing the robot and object to be manipulated even in manual mode. The image obtained by the camera can be adjusted by software using the implemented tools before the automatic manipulation begins, allowing the user to set the correct input parameters to ensure reliable object identification. Conclusions. In order to design a robotic system whose operation is controlled on the basis of visual perception, it was necessary to acquire theoretical knowledge for the correct selection of individual components of the system as well as their correct placement within the robotic cell. Great emphasis was placed on suitable and economical selection of the sensing device and the way of illuminating the scanned objects. In order to obtain the camera image it was also necessary to study and understand the principle of working with the image captured by the camera via SDK issued directly by the camera manufacturer. However, obtaining an image was only the first step to start the image processing process. In order to extract the necessary data from the obtained image and then to create control instructions from the data for controlling the robot, it was necessary to study and learn in detail the individual steps and procedures of image processing. In the part of the work dealing with image processing the acquired knowledge was applied to the processing process itself, but not only known approaches were used. Owing to the reduction of CPU load and consequently shortening of the calculation process, own procedures were also introduced into the image processing process. The actual "economical" approach was applied and tested in the thresholding process where a "shortened thresholding algorithm" was created. The approach was also applied to the object-in-picture search process, creating a "network-based object-in-picture method" that uses the fact that wesearch and identify known objects in industrial applications as opposed to identifying objects in an unknown environment. The combination of image acquisition, image processing and robot control with one comprehensive application is also a majorbenefit. Of course, to ensure this functionality, it was first necessary to create a theoretical base on which to build. The main problem was to create a control part of the robot control in C # and to link it to the basic control program created in C ++.