Davis is interested in computer vision and machine learning and his research focuses on modeling 3D motion to enable intelligent systems that understand dynamic 3D objects, humans, and scenes. He aims to develop this dynamic understanding by solving challenging problems in perception and prediction from videos and point clouds, such as 3D pose estimation, shape reconstruction, and motion forecasting. His work leverages learned and physics-based models of dynamics to incorporate a higher-level understanding of how 3D entities move and interact with their surroundings.