BREGENZ, Austria -- An innovative new design tool is due for immediate release by the Austrian-based design and IP company Photeon Technologies. The Optical Design Automation(tm) Tool uses advanced artificial intelligence (AI) to help find the global optimum for design problems.

Potential applications wide-ranging

While the software was initially designed by Photeon's engineers to facilitate the design of components for planar lightwave circuits, the potential applications are manifold. The spectrum extends from the design and optimization of optical components and lenses, to optimizing network routing for trunk traffic and assisting automobile traffic management in large cities. The tool is also intended for use by the lighting industry for the optimization of lighting solutions, as well as the automotive industry for motor management.

Patent-pending

A combination of genetic algorithms and the advanced learning ability of artificial neural networks enables the ODA(tm) Tool to find the global maximum of a problem more efficiently than other processes. A patent application has been filed for the software, which has already duplicated two existing patents and successfully developed a number of new design processes.

Genetic algorithms find the global maximum

Particularly suitable for design processes, genetic programming is increasingly being used in a number of areas from designing antennas to general-purpose controllers. In contrast to humans, artificial intelligence is unfettered by underlying mathematical knowledge and simply aims to produce improved results. The ODA(tm) Tool applies and repeats three "evolutionary" steps - creation of the next generation, crossover, and mutation - until the optimal solution is found. Instead of eliminating "rejects", the tool simply reduces the probability of their reproducing, thereby ensuring the gene pool does not become too limited and enabling the global maximum, not just local maxima, to be found.

Artificial neural networks - learning from experience

Simulating the brain's learning and information-processing functions, an artificial neural network allows the ODA(tm) Tool to learn from its own experiences, as well as from the results computed by the genetic algorithms. Using neural networks, instead of storing all the collected data within a database, also means calculation times remain constant, with the added advantage that the network can recall correct results for a problem for which it has been trained, even if the problem to be solved is not exactly the same as ones solved previously.

Minimal human input

Once the user has entered the initial data, the tool takes over and leaves the engineer free to work on other tasks. No further input is required, nor does the engineer need to wait until the optimization is complete to have an approximate solution. Man hours are reduced, resulting in lower development costs and shorter time-to-market.

Photeon Technologies GmbH