Razvoj programske podrške za robotski 3D strojni vid

master's thesis

• Title (english): Development of a software framework for robotic 3D machine vision
• Author: Marko Čutura
• Mentor: Marko Švaco (mentor)
• Mentor: Filip Šuligoj (komentor)
• Granter: University of Zagreb; Faculty of Mechanical Engineering and Naval Architecture; Zagreb
• Defense date and country: 2022-05-12, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Electrical Engineering; Automation and Robotics

Abstract

U ovom radu izrađen je sustav za lokalizaciju predmeta u nestrukturiranoj okolini. Implementiran je sustav koji se sastoji od Zivid 3D stereo kamere koju se može koristiti kroz Zivid Studio. U svrhu izrade sustava Zivid kamera implementirana je za online primjenu pomoću njezine C++ biblioteke koja dolazi od strane proizvođača. Uz kameru implementiran je algoritam za lokalizaciju baziran na prepoznavanju pomoću C++ PCL(Point Cloud Library) biblioteke. Korišteni su koncepti modela i scene, gdje je model objekt poznate pozicije i orijentacije, a scena predstavlja nestrukturiranu okolinu na kojoj se traži objekt ili komad od interesa, pomoću algoritma za lokalizaciju nalazi se objekt na sceni pomoću već postojećeg modela. Također, u radu je implementiran robotski manipulator Yaskawa CSDA10F uparen sa kontrolerom FS100 koji izvršava pozicioniranje TCP(Tool Center Point) alata u odnosu na lokalizirani objekt čime se potvrđuje uspješnost lokalizacije. Da bi se robot mogao implementirati napravljena je Hand Eye kalibracija robota sa Zivid kamerom, koja povezuje bazni koordinatni sustav robota sa ishodištem kamere. Napravljena je C aplikacija u Motoplus biblioteci koja omogućava sljedeće opcije potrebne za funkcioniranje rada: komunikaciju robota sa računalom pomoću Socket komunikacije, funkcije za transformaciju rezultata algoritama lokalizacije u pozicijsku varijablu robota i funkcije za manipulaciju ostalim varijablama i signalima robota. Da bi se prikazala praktična primjena navedenog rada, izrađena je Bin Picking aplikacija u kojoj robot izuzima predmet pronađen pomoću sustava za lokalizaciju i odlaže ga na unaprijed predviđeno mjesto. Za izuzimanje i manipulaciju predmetom konstruirana je i izrađena vakuumska prihvatnica.

Abstract (english)

In this thesis a system for localization of pieces in unstructured environment was developed. A system is implemented that consists of Zivid 3D stereo camera which can be used through Zivid Studio. For the purpose of building this system Zivid camera was implemented in an online fashion, using its own C++ Library supplied by the manufacturer. Alongside the camera an algorithm for localization based on recognition was implemented using the C++ PCL(Point Cloud Library) library. Concepts of model and scene were used, where the model stands for an object of known position and scene stands for an unstructured environment, using the localization algorithm it is possible to find an object on scene using an already known model of object. Also, in this thesis a robot manipulator Yaskawa CSDA10F was implemented paired up with the controller FS100, which executes the positioning of its TCP(Tool Center Point) in reference to the localized object, which confirms the success of localization. To implement the robot, a Hand Eye calibration procedure between the robot and the Zivid Camera was carried out, which connects the base coordinate system of the robot with the center of the camera. A C language application was created using the Motoplus library which allows the use of following options needed for system to function properly: communication of robot with the computer using the socket communication, functions for transformation of the results given by the localization algorithm to the position variables of the robot, and lastly, functions for manipulation with other robot variables and signals. To demonstrate the practical use of the aforementioned work, a Bin Picking application was developed in which a robot picks up the piece found using the system for localization and places it on a predetermined location. For the purpose of picking a vacuum gripper was designed and fabricated.

• Keywords: stereo kamera; matrica homogenih transformacija; Yaskawa; CSDA10F; FS100; Motoplus; lokalizacija; prepoznavanje; C; C++; Point Cloud Library; Zivid; Hand Eye kalibracija; Bin Picking; vakuumska prihvatnica
• Keywords (english): stereo camera; homogenous transformation matrix; Yaskawa; CSDA10F; FS100; Motoplus; localization; recognition; C; C++; Point Cloud Library; Zivid; Hand Eye calibration; Bin Picking; vacuum gripper
• Language: croatian
• URN:NBN: urn:nbn:hr:235:775377
• Study programme: Title: Mechanical Engineering; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka strojarstva (magistar/magistra inženjer/inženjerka strojarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2022-05-05 09:11:57