Integracija robotskega sistema za manipulacijo predmetov na tekočem traku z uporabo strojnega vida

Kveder, Brina (avtor), Tomažič, Simon (mentor)

Pobiranje in razvrščanje predmetov s premikajočega se tekočega traku predstavlja pogost problem v industrijski avtomatizaciji, saj zahteva natančno sinhronizacijo med zaznavo, obdelavo podatkov in robotsko manipulacijo v realnem času. Diplomsko delo obravnava razvoj in integracijo vizijsko vodenega robotskega sistema za zaznavo, pobiranje in razvrščanje kock na tekočem traku. Sistem sestavljajo pametna industrijska kamera Zebra VS70, programirljivi logični krmilnik (PLK), tekoči trak z enkoderjem ter dva industrijska robota: 6- osni robot in robot SCARA. 6-osni robot pripravlja tok predmetov z odlaganjem kock na tekoči trak, robot SCARA pa s pomočjo funkcije za sledenje tekočemu traku (angl. conveyor tracking) sledi premikajočim se predmetom, jih pobira ter razvršča glede na rezultat razpoznavanja s kamero. V okviru naloge so bile uporabljene metode strojnega vida za zaznavo položaja in klasifikacijo kock, izvedena je bila kalibracija koordinatnih sistemov ter vzpostavljena komunikacija med komponentami prek industrijskega protokola EtherNet/IP. Razvita je bila tudi programska logika za PLK in robota, ki omogoča sinhronizirano delovanje celotnega sistema. Uspešnost sistema je bila ovrednotena z analizo pravilnosti štetja pik, uspešnosti pobiranja in odlaganja, latence, takta sistema, prepustnosti ter operativne zanesljivosti. Rezultati so pokazali 97-odstotno uspešnost pravilnega štetja pik, 92-odstotno pogojno uspešnost pobiranja in 96,7-odstotno pogojno uspešnost odlaganja. Celotna uspešnost zaključenega procesa je znašala 89 odstotkov. Latenca posamezne kocke je bila med približno 8 s in 9,7 s, ocenjeni takt sistema pa približno 7,4 s. Dejanska prepustnost sistema je znašala približno 7,11 uspešno zaključenih kock na minuto. Rezultati potrjujejo ustrezno delovanje sistema za demonstracijsko uporabo. Glavne omejitve sistema predstavljajo mehanske lastnosti prijemala ter zmanjšana zanesljivost pri višjih hitrostih tekočega traku.

• Jezik: Slovenski jezik
• Ključne besede: strojni vid, robot SCARA, 6-osni robot, sledenje tekočemu traku, programirljivi logični krmilnik, EtherNet/IP, industrijska avtomatizacija
• Vrsta gradiva: Diplomsko delo/naloga
• Organizacija: FE - Fakulteta za elektrotehniko
• Leto izida: 2026
• PID: 20.500.12556/RUL-186926
• Datum objave v RUL: 07.09.2026

Sekundarni jezik

• Jezik: Angleški jezik
• Naslov: Integration of a Robotic System for Conveyor-Based Object Manipulation Using Machine Vision

Izvleček

Picking and sorting objects from a moving conveyor belt is a common challenge in industrial automation, as it requires precise synchronization between object detection, data processing, and robotic manipulation in real time. This thesis addresses the development and integration of a vision-guided robotic system for the detection, picking, and sorting of dice on a conveyor belt. The system consists of a Zebra VS70 smart industrial camera, a programmable logic controller (PLC), a conveyor belt equipped with an encoder, and two industrial robots: a 6-axis robot and a SCARA robot. The 6-axis robot prepares the object flow by placing dice onto the conveyor belt, while the SCARA robot uses the ’conveyor tracking’ function to follow moving objects, pick them, and sort them according to the recognition results provided by the camera. Within the scope of this thesis, machine vision methods were used for object position detection and dice classification. Calibration of coordinate systems was performed, and communication between system components was established via the industrial EtherNet/IP protocol. Control logic for both the PLC and the robots was also developed to ensure synchronized operation of the entire system. System performance was evaluated by analyzing the accuracy of pip counting, picking and placing success, latency, system takt time, throughput, and operational reliability. The results showed a 97-percent success rate in correct pip counting, a 92-percent conditional picking success rate, and a 96,7-percent conditional placing success rate. The overall success rate of the completed process was 89 percent. The latency of an individual die ranged from approximately 8 s to 9.7 s, while the estimated system takt time was approximately 7.4 s. The actual system throughput was approximately 7.11 successfully completed dice per minute. The results confirm that the system is suitable for demonstrational purposes. The main limitations of the system are the mechanical characteristics of the gripper and reduced reliability at higher conveyor belt speeds.

• Ključne besede: machine vision, SCARA robot, 6-axis robot, conveyor tracking, programmable logic controller, EtherNet/IP, industrial automation