Research on the influence of robotic manipulation on the uncertainty of product dimension measurements

Zore, Aleš (author), Munih, Marko (mentor)

In this doctoral thesis, we present research on the influence of robotic manipulation on the uncertainty of product dimension measurements. We have been studying a robotic measurement cell dedicated to statistical process measurements. By robotizing process measurements, the robot becomes part of the measurement system, and the properties of the robot directly influence the measurement process. As part of the PhD thesis, the robot and robot manipulation were considered influential elements in the uncertainty of object dimension measurements. The first chapter provides insight into statistical process measurements currently used in the process industry. Performance criteria for measurement and process control systems are presented. The introductory chapter concludes with the current state of the art of robotic process measurement and the fundamental issues of the thesis. Chapter 2 presents the experimental system used – a robotic cell for statistical process control. It highlights the key features of the UR5e collaborative robot and the Renishaw Equator 300 comparative principle based measurement robot (PME). The chapter describes the two geometrically and dimensionally different types of serial products considered. The six most critical dimensional characteristics in the daily statistical process control are followed by a description of the implementation and evaluation of the measurement system analysis (MSA) procedures. The most common MSA outputs are the capability parameter Cg, the standard deviation and the range of measurements R in the case of the evaluation of the capability of a gauge, and the repeatability and reproducibility parameter GRR in the case of the verification of the capability of a measurement process. This is followed by a theoretical demonstration of measurement uncertainty and metrological traceability in practice. Chapter 3 of the thesis addresses the initial question of whether robot manipula- tion impacts dimensional measurements. Nine different manipulation scenarios, ran- ging from zero manipulation to full manipulation, used the robot as a manipulator to carry and insert measurement objects into the PME. Each robot manipulation scena- rio consists of three measurement runs with 25 repetitions, each of which is a robot manipulation and PME dimensional measurement. The effect of robot manipulation is discussed based on procedure 1 MSA statistical analysis. In addition to addressing the complexities of robotic manipulation, the mode of operation (discrete touch and scan) and PME sampling were addressed separately. As the complexity of the robotic manipulation increases (adding longer movements, larger changes in orientation, and multiple grasps), the scatter of the measurements increases. However, no difference in scatter was detectable for the two modes of operation and for different numbers of PME sampling points. In Chapter 4, special attention is paid to the trajectory of the robot’s end-effector. The way of interpolating the robot’s motion (linear or joint motion), the different ve- locities of the motion, the length of the motion, and the number of active joints to perform the robot’s motion are highlighted. The influence on the accuracy and preci- sion of the robot tip is identified. The error in the accuracy and precision of the robot tip is detected indirectly through the variability of the dimension measurements. The more significant variability of the dimension measurements is due to the variability of the position of the inserted measurement object in the fixture. The captured dimensio- nal data were analysed using the MSA procedure 1, and the statistical independencies of the individual influential trajectory parameters are further addressed using the ana- lysis of variance (ANOVA) method. Each exposed parameter has at least a minimal influence on the robot’s accuracy and precision or on the uncertainty of the dimensional measurements. The last chapter focuses on the robotic grasping operation. An optical measurement system with two line laser sensors is designed to observe the change of orientation of an object. In addition to the dimensional measurements, the differences in all three orientation angles are observed when the robot is only touched and released. The rotation about the vertical axis of the axisymmetric object is more pronounced in the self-touch comparison. The rotation about the other two axes is not pronounced. Considering the variation between touch and release, the differences in the perceived twist between the different applied gripping forces are practically negligible.

• Language: English
• Keywords: robot, process measurement, measurement uncertainty, dimensional mea- surement, comparative measurement, robot manipulation, measurement robotization
• Work type: Doctoral dissertation
• Organization: FE - Faculty of Electrical Engineering
• Year: 2024
• PID: 20.500.12556/RUL-162465
• Publication date in RUL: 24.09.2024

Secondary language

• Language: Slovenian
• Title: Raziskava vpliva manipulacije z robotom na negotovost meritev dimenzij izdelkov

Abstract

V doktorski disertaciji predstavljamo raziskavo vpliva manipulacije z robotom na izmerjeno negotovost dimenzij izdelkov. Delo obravnava robotsko merilno celico namenjeno statističnim procesnim meritvam. Z robotizacijo meritev v proizvodnih procesih robot postane del merilnega sistema in njegove lastnosti neposredno vplivajo na proces meritev. V sklopu doktorske disertacije je robot ter manipulacija z robotom obravnavana kot vplivni element na negotovost meritev dimenzij objektov. Prvo poglavje predstavlja vpogled v statistične procesne meritve trenutno uporabljene v kosovni proizvodnji. Prikazani so kriteriji zmogljivosti merilnih sistemov in sistemov za vodenje proizvodnega procesa. Trenutno stanje robotizacije proizvodnih meritev in temeljna vprašanja disertacije zaključujejo uvodno poglavje. Drugo poglavje predstavlja uporabljen eksperimentalni sistem – robotsko celico namenjeno statistični procesni kontroli. Poudarjene so ključne lastnosti uporabljenega sodelujočega robota UR5e in primerjalno-merilnega robot Renishav Equator 300 (PME). Poglavje se nadaljuje z opisom dveh obravnavanih geometrijsko in dimenzijsko različnih tipov serijskih produktov. Šestim najbolj kritičnim dimenzijskim karakteristikam, vsebovanih v vsakodnevno statistično procesno kontrolo, sledi opis izvajanja in vrednotenja procedur analize merilnega sistema (MSA). Najbolj pogosti rezultati MSA so parameter sposobnosti Cg, standardni odklon in razpon meritev R v primeru vrednotenja sposobnosti merila ter parameter ponovljivosti in primerljivosti GRR (ang. repeatability and reproducibility) pri verifikaciji sposobnosti merilnega procesa. Sledi teoretični prikaz merilne negotovosti in meroslovne sledljivosti v praksi. Izhodiščno vprašanje, ali ima manipulacija z robotom vpliv na dimenzijske meritve, je obravnavano v tretjem poglavju disertacije. V definiranih devetih različnih scenarijih manipulacije z robotom, od nič manipulacije do polne robotske manipulacije, je bil robot uporabljen kot manipulator za prenašanje in vstavljanje merilnih objektov v PME. Posamezen scenarij manipulacije z robotom vsebuje tri serije meritev s po 25 ponovitvami manipulacije z robotom in dimenzijskih meritev s PME. Vpliv manipulacije z robotom je obravnavan na podlagi procedure 1 MSA statistične analize. Poleg obravnave kompleksnosti robotske manipulacije sta bila ločeno obravnava način delovanja (diskretno dotikanje in skeniranje) in vzorčenja PME. Z višanjem kompleksnosti robotske manipulacije (dodajanje daljših gibov, večje spremembe v orientaciji in večkratna prijemanja) raztros meritev narašča, medtem ko razlika v raztrosu pri dveh načinih delovanja in pri različnemu številu vzorčnih točk PME ni bila zaznavna. V četrtem poglavju je posebna pozornost usmerjena v obravnavo trajektorije vrha robota. Izpostavljen je način interpolacije giba robota (linearen gib oziroma gib po sklepih), različne hitrosti giba, dolžina giba in število aktivnih sklepov za izvedbo robotskega giba. Prepoznan je vpliv na točnost in natančnost vrha robota. Napaka točnosti in natančnosti vrha robota je zaznana posredno preko raztrosa meritev dimenzij, večji raztros meritev dimenzij je posledica variabilnosti lege vstavljenega merilnega objekta v ležišču merilnika. Zajeti dimenzijski podatki so analizirani s proceduro 1 analize MSA, statistične neodvisnosti posameznih vplivnih parametrov trajektorije pa so nadalje obravnavane z metodo analize variance (ANOVA). Vsak izpostavljen para- meter ima vsaj minimalen vpliv na točnost in natančnost robota oziroma na negotovost meritev dimenzij. Zadnje poglavje je namenjeno obravnavi operaciji robotskega prijemanja. Za opazovanje spremembe orientacije objekta je zasnovan optični merilni sistem z dvema linijskima laserskima senzorjema. Poleg dimenzijskih meritev so opazovane razlike v vseh treh kotih orientacije pri samo dotiku in izpenjanju z robotom. Zasuk okoli vertikalne osi osno-simetričnega objekta je izrazitejši izpenjanju s primerjavo samo dotika. Zasuka okoli preostalih dveh osi nista izrazita. Glede na spremembe med dotikom in izpenjanjem so razlike v zaznanih zasukih med različnimi apliciranimi silami prijemanja praktično zanemarljive.

• Keywords: robot, proizvodne meritve, merilna negotovost, merjenje dimenzij, primerjalne meritve, manipulacija z robotom, robotizacija meritev