Caleb's thesis goals are to incorporate artificial intelligence with modern deep learning techniques in the design of optical components to develop high level expert systems bringing optical design into an easy to use format where anyone can create high performing systems.

Specifically, Caleb hopes his efforts will provide: 1.) An AI for illumination lenses capable of reproducing the irradiance pattern of any input image for a given geometry. 2.) An AI for Imaging systems that can generate an optical system with the desired performance given mounting and space limitations. 3.) Generalization of the methods learned from these techniques to other applications within optical design, such as adaptive optics and diffractive grating design, in addition to investigating unsupervised learning techniques for AI self-training in these areas.

Update 2020

I am currently investigating methods for ultrafast ray-tracing of freeform optical surfaces defined using NURBS, and algorithms for optimizing those surfaces which take advantage of the accelerated speeds. I am also heavily exploring AI/generative algorithms for creating art, which I expect to provide insight for AI driven optical design in the future