Action is indelibly tied to perception, and good perception is vital to our survival. Action helps to inform organisms about the world they inhabit, and to accrue information that aids in achieving their goals in a timely manner. In many computer vision applications, action plays no role in understanding an image; yet in natural situations, action and perception are inextricably linked across all levels. My thesis talk is about building machines that operate in real time in the real world, using actions to aid perception. I will present many examples of active perception domains, each with challenging problems. We analyze these problems mathematically, propose solutions, and evaluate their performance in real world conditions. In the process, we explore a robust mathematical foundation for understanding active perception, and develop many techniques for practical analysis.