Recognition and sorting of colored objects with machine vision and a robot

Zaključna naloga obravnava robotski sistem za pobiranje, strego in odlaganje barvnih kock iz zalogovnika do odlagalnega mesta in obratno. Prvi del naloge obravnava teoretične osnove robotike in strojnega vida. Predstavljene so različne vrste robotov in robotskih prijemal. Nato so predstavljene osnove strojnega vida in obdelave ter analize slik. V metodologiji so opisane vse komponente obravnavanega sistema in idejna zasnova poteka operacij strežnih operacij in operacij spusti in primi. V eksperimentalnem delu smo za operacije strege in primi in spusti analizirali in ocenili ponovljivost in natančnost robotskega sistema. Del rezultatov predstavlja tudi programska koda za vse robotske operacije. Programi so razdeljeni na del strege in operacij z robotom ter del operacij, ki jih izvajamo s strojnim vidom, to je zajem, obdelava in analiza slike. Ker vseh preizkusov za vse lokacije v skladišču in odlagalnem mestu nismo izvedli, je predstavljen tudi način računskega določanja manjkajočih lokacij. Na ta način lahko popišemo vse lokacije koordinatnih sistemov posameznih podsistemov in vse lokacije znotraj podsistemov. S pridobljenimi podatki lahko enostavno časovno ovrednotimo vse strežne robotske operacije vnaprej in izdelamo nove programske kode.The thesis addresses a robotic system for pick-and-place operation and handling of colored blocks from a storage container to a designated location and vice versa. The first part of the thesis covers the theoretical fundamentals of robotics and machine vision. Various types of robots and robotic grippers are presented. Subsequently, the basics of machine vision, image processing, and analysis are introduced. The methodology describes all components of the considered system and outlines the conceptual design of the pick-and-place operations. In the experimental section, the repeatability and precision of the robotic system operations are analyzed and assessed. A portion of the results also includes the source code for all robotic operations. The programs are divided into segments for the pick-and-place and handling robotic operations, as well as the machine vison operations - namely, image capturing, processing, and analysis. As we did not conduct tests for all pick-and-place locations within the warehouse and the depositing area, a method for calculating the missing locations is used. This approach allows us to determine the coordinate system locations of individual subsystems and all the locations within these subsystems. With the acquired data, we can readily evaluate the time requirements for all robotic operations in advance and generate new sets of program codes