The goal of our diploma thesis consisted of tracking and picking a moving object with the 'AdeptOne' SCARA-Robot. To reach this goal the previously used image processing system was replaced by a faster one. Another part of this work was to become acquainted with the ADEPT-System and the new image processing system. The track and pick algorithm had to be as independent as possible of the speed and direction of the object.

The new image processing system consisted of the 'Cognachrome Vision System' from 'Newton Research Labs' and a SONY colour-camera. The properties of this system are a fast object recognition (60 evaluated pictures per second) and the ability to recognise previously trained colour-areas. The object-shape is of no importance for the recognition.

The installation of the camera and 'Vision-Board' on the robot-arm made a tracking and picking of the object possible over the entire robot-workspace. Further, by using a short video-cable the image quality was maintained high. Arranging the correct light conditions and choosing the most suitable background and object colours we were able to obtain a proper recognition of the object over the whole workspace. The camera was calibrated with its lens looking beneath the robot-tool.

Both 'Cognachrome Vision System' and ADEPT-System were connected by a PC via the serial ports. The image-data is read and error-checked by the PC and then it is sent to the ADEPT-System where the last step of processing for a robot-motion is made. The time needed by these process-sequences to be accomplished is determined solely by the ADEPT-System. A relief of this system was obtained by processing the image-data on the PC and by the convenient camera calibration.

The track and pick algorithms developed by us are able to follow and catch an object moving within a speed-range of 0,2 to 0,5 m/s.

The diploma work is based on a automatic robot measuring system. The work was initiated by the firm Innotest AG because they wanted to automatise the time consuming part of the calibration of the air ultrasonic measuring instrument.

The measuring station includes: The air ultrasonic measuring instrument used to measure the thickness of the powder layer. The measuring instrument was developed by Innotest AG and partners. There is also a Kawasaki six axes folding arm robot that was provided for the time of the diploma work by Kaiser engineering AG. To control the different signals we used a I/O card that is installed in a normal PC. The software that controls the card was written with the development tool Visual C++.

The job of the robot is to move the ultrasonic sensor over a aluminium plate covered with powder. The Data from the measuring instrument are used in the control software to create the signals for the robot controller. With this signals the robot is able to find the optimal distance and through a measuring routine which we worked out it can choose the optimal angle.

On the basis of the measuring results we got from tests on clean aluminium plates the specialists of the Innotest AG could confirm that we are on the right way with our measuring routine. In the last week of the diploma work we had the opportunity to test our measuring station directly at the Innotest AG. This tests we did with the original powder-coated metal plates. The results of this tests were compared to the results produced by hand measuring. The results proved to be almost the same.

One of the projects of the Institute of Mechatronic Systems IMS is to develop a Java based Robotics Component Framework RCF, which will standardize and simplify the development of robot control system software. To research and test this software, a robot with particular specifications was needed.

The first task of the team was to develop and build the robot. Secondly, the robot had to be put into operation by using the realtime operating system Jbed and a prototype of a RCF based Controller Software.

During this project, the team worked extensively in the fields of mechanical engineering, electronics and computer science, where the following steps had to be fulfilled: A robotarm from the Institute of Robotics in Zurich was the base for the design and construction of a four degrees of freedom robot. Parallel to this construcion, schematics of a control panel had to be utilized, the panel itself assembled and tested. Then the prototype of the RCF based Controller Software was designed, some components were implemented, but the Controller Software itself could ultimately not be put into operation because until the end of this project no functional version of Jbed was available.

As the robot stands now, it presents a valid base for further research in the field of Java based Robotics Component Framework.

The development of ink jet systems is of significant economical importance. Since mass market is dominated by ink jet printers there is a big interest in manufacturing better printers with improved performance. Of particular concern is the miniaturization of bubble jet nozzles that can keep up with high performance standards.

To reach the goals there is no way around simulation. Modeling and simulating the physics that goes on in the bubble jet nozzle is the central theme of this thesis. The analysis usually goes by six steps: heating, nucleation, the growing of the nuclei, the formation of the liquid film, the growing of the film and the fluid dynamics. This complexity requires each step in the simulation process to rest on firm grounds. Experimental and numerical results reported in the current literature still come with great uncertainties. Often it is unclear how these experiments where done and to what extent they can be used in a manufacturing process. The purpose of this diploma thesis was the validation of some of the results that concern the heating process.

We used three methods to analyze the 3D heat conduction that takes place when the nozzle heats up. The first model is a distributed parameter model that has been solved numerically using the finite element software package NM SESES. The other models are based on an analytic ansatz and a lumped parameter model, respectively. Since these two models do not give satisfactory results in 3D we reduced the system to one dimension. The results we got when applying the three methods to this 1D system have been compared with experimental results and analytical solutions.

Our results using finite element simulation confirm results obtained in earlier research where finite difference algorithms have been applied. Our results diminish the above mentioned uncertainties and add confidence in numerical simulations of bubble jet nozzles.