Real-time GNSS positioning based virtual working gauge for serial robotic manipulator
Mäkinen, Olli (2021)
2021
Konetekniikan DI-ohjelma - Master's Programme in Mechanical Engineering
Tekniikan ja luonnontieteiden tiedekunta - Faculty of Engineering and Natural Sciences
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2021-08-12
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202107266350
https://urn.fi/URN:NBN:fi:tuni-202107266350
Tiivistelmä
In this thesis a control system that could limit robot manipulators reach when its global position was know was developed. The reach was limited to a virtual working gauge using global navigation satellite system data on the manipulators position and a building information model of the environment. The control system was designed such that these limits would not be violated. Functionality of the control system was simulated in MATLAB Simulink environment.
Building information model and satellite positioning can be used to infer different infrastructure elements exact location in relation to the manipulator. If the manipulator was to move, the global navigation satellite system can sense that and update its relative position in regards to the building information model. Utilising building information model is also especially useful in defining virtual limits for features for which their exact location is hard to define using traditional perception methods. To maintain manipulator movements inside these defined limits, the control system will calculate the shortest distance from the manipulator to the limits. Defining exact distances between complex shapes is resource intensive and therefore the manipulator representation was simplified to each of its links bounding boxes. This way the distance calculations could be ensured to work in a real time system. Two control methods were implemented for modifying the manipulator joint speeds based on the distance information. Manipulator movements could be slowed down linearly when approaching a limit or alternatively using an artificial potential field function to generate a path towards the target while repelling the manipulator away from the limits.
As an case study the thesis researched defining virtual working gauge limits for an road rail excavator that operates close to an train track open for traffic. In this situation the road rail excavators reach should be limited such that no part of it crosses the working gauge limits of the nearby train track. Building information model is suitable in this context as defining the exact location of the train track using perception methods is difficult.
Creating and maintaining virtual working gauge limits using the control system developed was deemed possible. For infrastructure elements that are normally expressed as just lines of connected points in current building information models need to be adjusted by creating bounding volumes around them to be usable in a virtual working gauge application. The ways in which the control system modifies the manipulator movements work as desired in the simulation environment. However the manipulator representation as bounding boxes can lead to singularities, where there are multiple options for the location of the shortest distance. Ways of calculating multiple shortest distances along the manipulator bounding boxes need to be utilised to eliminate slight vibrations in the controller desired velocity signal that are caused by singularities when calculating only single shortest distance per bounding box.
Building information model and satellite positioning can be used to infer different infrastructure elements exact location in relation to the manipulator. If the manipulator was to move, the global navigation satellite system can sense that and update its relative position in regards to the building information model. Utilising building information model is also especially useful in defining virtual limits for features for which their exact location is hard to define using traditional perception methods. To maintain manipulator movements inside these defined limits, the control system will calculate the shortest distance from the manipulator to the limits. Defining exact distances between complex shapes is resource intensive and therefore the manipulator representation was simplified to each of its links bounding boxes. This way the distance calculations could be ensured to work in a real time system. Two control methods were implemented for modifying the manipulator joint speeds based on the distance information. Manipulator movements could be slowed down linearly when approaching a limit or alternatively using an artificial potential field function to generate a path towards the target while repelling the manipulator away from the limits.
As an case study the thesis researched defining virtual working gauge limits for an road rail excavator that operates close to an train track open for traffic. In this situation the road rail excavators reach should be limited such that no part of it crosses the working gauge limits of the nearby train track. Building information model is suitable in this context as defining the exact location of the train track using perception methods is difficult.
Creating and maintaining virtual working gauge limits using the control system developed was deemed possible. For infrastructure elements that are normally expressed as just lines of connected points in current building information models need to be adjusted by creating bounding volumes around them to be usable in a virtual working gauge application. The ways in which the control system modifies the manipulator movements work as desired in the simulation environment. However the manipulator representation as bounding boxes can lead to singularities, where there are multiple options for the location of the shortest distance. Ways of calculating multiple shortest distances along the manipulator bounding boxes need to be utilised to eliminate slight vibrations in the controller desired velocity signal that are caused by singularities when calculating only single shortest distance per bounding box.