Fast approximative data structures for applications in the automotive industry
Date issued
Authors
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
License
Abstract
In the automotive industry digital prototyping becomes more and more important. By digital prototypes many tests can be done virtually, the digital design process can be adapted dynamically and thus costs can be reduced. The basic algorithms used in this context are always the same and mainly concern the calculation of distances and penetrations.
In this work an approach will be presented how distances and penetrations can be calculated approximately very fast. This approach will be applied to pack arbitrarily shaped objects into a trunk, which is one of many problems arising in the digital design process. Because the geometry describing a trunk is usually a “soup” of triangles which might even contain large holes, at first a robust heuristic will be presented how for such input data a voxel representation of the interior of a trunk can be computed.
Another field where distance calculations play an important role are checks concerning the engine. By data generated during a test drive and consisting of rigid transformations describing the vibration movement of an engine it can be checked for the digital model of the engine that a certain safety distance to other parts is maintained.
A new approach will be presented how the concept of bounding volumes, which is only applied to 3-dimensional geometry so far, can be transferred to rigid transformations. By this approach the minimum distance between the engine and the surrounding parts, that occurs throughout a test drive, can be calculated very fast. Furthermore, by this approach the course of the distance between engine and surrounding parts during the test drive can be calculated approximately very fast as well.
The motion data of an engine during a test drive is collected by motion tracking systems and several types of such systems have become available at low costs in the last years. Therefore, finally an approach will be shown how data collected by several motion tracking systems fixed to the same object can be combined and put into geometric relation to each other allowing a comparison of the recorded data.