Abstract: Light is an electromagnetic spectrum that spans a wide range of wavelengths, including ultraviolet, visible, infrared, thermal wavelengths, and so on. Human vision is restricted to the visible wavelengths, but light interacts a lot more richly in the infrared and thermal wavelengths. Moreover, thermal images are clear even in harsh weather conditions like rain, and fog. In this talk, I will present three ideas towards robust imaging with thermal cameras. First, I will talk about combining the physics of thermal image formation with deep learning to enhance the resolution and reduce noise that is endemic to thermal cameras. Second, I will show how thermodynamics can be leveraged to perform robust material classification. Finally, I will talk about how new types of optics, called meta-optics, can be leveraged to dramatically reduce the size of thermal cameras, while enhancing their function.