Shape Sensing using Modal Transformation Theory and Neural Networking 
NASA Dryden in Edwards Air Force Base in California solicited proposals for undergraduate students to compete for 4 summer internship positions. The general solicitation topic was dynamic flight loads. The winners received an 11 week paid internship at the center, to explore their proposed research topic.
It has been shown that for simple structures such as a 2D cantilever beam problem, the first four natural mode shapes and as few as four discrete strain values can be combined to very accurately predict the entire shape of the beam. This strategy is very appealing because if the mode shapes of a structure are known, then that structure's shape can be monitored in real time using simple strain gauges.
The challenge for expanding this procedure to more complex shapes is that the mathematics become prohibitively complex and in most cases there is not a closed form solution like that of the 2D cantilever beam. Therefore, an advanced neural network was introduced that could use the known mode shapes as well as a series of on-the-ground training data and automatically create weighted coefficients. These values act as a conversion factor, allowing discrete strain measurements to be quickly extrapolated to the entire deformed shape.