Force Control and Constraint-based Task Specification in Robotic Assembly

Abstract

Små- til mellomstore bedrifter har utfordringer med å automatisere deres produksjon grunnet kostnader tilknyttet lite fleksible roboter. For at roboter skal bli en lønnsom investering, så må de kunne gjennomføre ulike oppgaver i varierende omgivelser.

Denne masteroppgaven har som overordnet mål å utvikle et automatisk styringssystem for industriroboter, mer spesifikt ser denne oppgaven på hvordan eksterne krefter på arbeidsstykke kan benyttes til å styre roboten i ved bruk av en restriksjonsbasert tilnærming. Restriksjonsbasert robotstyring benytter seg av geometriske og dynamiske relasjoner mellom arbeidsstykke og omgivelser som veiledende styring av roboten.

Det er utviklet en kalibreringsrutine for en kraftsensoren og en "admittance" kraftstyring. Utviklingen er gjort i ROS og testet med simulueringer på en KUKA Agilus KR 6 industrirobot.

Det konkluderes med at videre arbeid burde utvikles i ROS2 grunnet nyere programvare og initialiseringsutfordringer fra denne oppgaven.

In order for a robot to be a profitable investment for small- and medium-sized enterprises, it should be able to execute a variety of tasks in partially structured or even unstructured environments.

This Master's thesis is aiming towards automating control of a manipulator in industries, more so, looks on the possibility of implementation of force/torque control in robotic assembly using constraint-based robot control. Constraint-based robot control uses the desired geometrical- and/or force-constraints of the workpiece to its environment as input to control the manipulator.

This is partially done by developing an admittance force control by manipulating the end-effector with second-order mass-spring-damper behavior. A calibration routine is developed. The work is simulated on a KUKA Agilus KR 6 manipulator and is integrated with ROS.

The programming is approached with the open-sourced Robot Operating System (ROS) and the framework for constraint-based robot programming, named eTaSL. The thesis aims to investigate the possibilities of force/torque control in robotic assembly tasks by using constraint-based robot control programming.

Measurements of forces and torques from the robot manipulator wrist is used as inputs in a robotic controller, which relates these to end-effector position in terms of a robot task that lead to desired control of the manipulator configuration.

It is shown how force inputs from an ATI Gamma IP60 force-torque sensor is calibrated and implemented as a constraint-based task in eTaSL to develop an admittance force control for KUKA KR6 R900.

It is concluded that force control can be implemented in constraint-based programming, more specific eTaSL. This is still in the early stage of development and presents an opportunity for further development and practical application. In further work, it is recommended to look into implementation of constraint-based robot control in ROS2 due to initialization issues and outdated software.