In fall 1994, I began my studies of »business engineering (Wirtschaftsingenieurwesen) at the »University of Karlsruhe. In the graduate courses, I focused on labor science and production technology. I have written both my one-term thesis and diploma thesis in the »institute for process control and robotics. Both theses focused on the problem of manipulating deformable linear objects by industrial robots.

Diploma thesis:

In the diploma thesis, the task was to analyze the economical efficiency of an assembly system based upon the HANDFLEX robot.

In the »HANDFLEX research project a robot system was developed that should handle deformable linear objects. The base was a »KR15 industrial robot from »KUKA equipped with various sensors. The HANDFLEX robot was planned to assemble cable harnesses to car doors. Even the economical analysis of an ordinary computer integrated manufacturing system has its problems, e.g. to find adequate solutions to handle non-monetary attributes like flexibility or product quality. In the case of HANDFLEX the need to extrapolate parameters from the prototype to the final machine poses additional difficulties.

The result was that an automated solution would be very efficient, if it could be made to work reliable and fast. However, problem was that the automated handling of deformable objects with sensors was quite slow and unreliable.

One-term thesis:

In the one-term thesis, the task was to research, if a model created with standard FEM Software could be used for supporting manipulations of deformable objects.

For many years, the manipulation of rigid objects by robots had been thoroughly studied and various systems had been realized with success. However, the manipulation of deformable objects is more difficult, as the objects change their shape during manipulation. An approach to solve this problem is to create a mathematical model of the behavior of the deformable object and to use this model to calculate the present shape of the object during manipulation. The modeling software used was »ABAQUS 5.7.

The result was that although the shape can be computed, it is not really helpful for manipulation because the parameter of the object have to be known exactly. In reality, these parameters are usually not known.