Zhengxiong, Yang
Zhengxiong Yang did his master's thesis at the Lehrstuhl für Bauinformatik at TU München in cooperation with Siemens AG. He graduated from come.tum/BGCE in 2006, and is now research assistant at the Lehrstuhl für Bauinformatik (see his homepage).
Master's Thesis:
Deterministic and probabilistic optimization of muti-mirrors
in HUD using RoDeO
A Head-Up Display, also known as a Heads-Up Display, or simply HUD, is any type of display that presents data without blocking the user's view. This technique was pioneered for military aviation and is now used in commercial aviation, motor vehicle and other applications. The automobile HUD projects information onto the windshield in a way that it appears to be floating in front of the driver. The virtual image is created through the use of optics and design. So, instead of having to glance down at the radio, the speedometer or the fuel gauge, drivers can look straight at the road and find that information.
During the manufacturing and assembly of HUD, the occurrence of optical deficiency caused by deformation of the concaved mirrors which are fixed onto the house can not be avoided completely due to manufacturing and assembly tolerances. In order to minimize these image distortions, a multi-mirror visualization system is built up to simulate the imaging process and to assess the quality of the image for a given geometry. With this model at hand, a probabilistic optimization of the geometry of the mirrors is performed to reduce the failure probability.
The probabilistic optimzation is of advantage when the input variables are of stochastic distribution. The deterministic method computes the minimum, because ue to the input tolerance, the output variation of the design point can be very great. To make the variation smaller, the new design point is greater but of smaller variation and mean value.
The probabilistic optimizer RoDeO (Robust design optimization) of SIEMENS AG, is enabled for the stochastic analysis based on Generalized Polynomial Chaos, FORM and SORM. With the help of RoDeO, both the variance and mean value of the image quanlity is minimized, i.e. the probability of failure is reduced.
