Here's a bit of what motivates my research: For autonomous systems to have general reasoning capabilities in a diverse and uncertain world, they must be able to reason about the future with rich perceptual input—can we design effective learning algorithms for agents to reason about and influence the future? Towards this question and others, I work on RL and IL methods at the interface of Computer Vision and Machine Learning. I'm specifically interested in decision-theoretic models that leverage rich perception sources to inform forecasting and control tasks. I'm also interested in creating systems that continually learn without extrinsic reward signals, and creating systems that cooperate reliably and intuitively with humans.