NASA Evolutionary Design
Bringing new designs to life…
The evolutionary process of natural selection proposes favorable traits become more common in successive generations of a population and unfavorable traits become less common.
Changes due to natural selection typically take millions of years to become evident. Imagine if this process could be expedited - where the strongest traits become immediately apparent. This is exactly what NASA researchers are doing in their Evolvable Systems Group.
Improving Traditional Design Techniques
NASA recently used evolutionary computation to design and optimize a satellite antenna to orbit earth. The software started with random antenna designs and refined them through evolutionary computation. Each element within a design was evaluated on its ability to perform a specified task. The strongest elements were placed together and reevaluated as the design “evolved”. This process continued until a superior device was produced. By using evolutionary computation, NASA created a broader range antenna that required less power and could fit into a one-inch space. NASA has shown that evolutionary computation holds great promise when compared to traditional design techniques which rely heavily on “trial and error.”
Increasing Efficiency
Methods of designing and optimizing antennas by hand are time and labor intensive, limit complexity and require significant expertise and experience. Although evolutionary computation is capable of creating innovative instrument designs that will be vital to future mission success, it does have some limitations. Evolutionary computation is very compute intensive and requires processing power typically found only in clusters or supercomputers. These resources are expensive and often in limited supply.
Limitless Possibilites
Through the Evolutionary Design initiative, NASA provided funding for the West Virginia High Technology Consortium (WVHTC) Foundation’s Global Grid Exchange® (G2EX) and Parabon® Computation, Inc. to make evolutionary computation grid-computing applicable. The resultant software development kit (SDK), entitled ORIGIN, provides a platform for creating applications that employ evolutionary computation algorithms designed to run atop large computational grids, such as G2EX. ORIGIN provides significant savings over traditional high compute-to-time solutions for researchers using evolutionary computation. Grid-based evolutionary design also allows the number of variables within a design to be virtually limitless and results to be returned with unmatched quickness.
Supporting Innovation
As NASA prepares for upcoming space missions, the need for innovative component designs has become vital. Future missions will experience extreme physical and environmental factors not yet encountered in space travel. Through the Evolutionary Design initiative, G2EX is providing NASA engineers the tools they need to create designs that extend
beyond human imagination.