Matt Deitke

Papers

Total Citations

H-Index

About

Matt Deitke is a leading researcher in embodied AI and computer vision, whose work bridges the critical gap between simulation and real-world robotics. His most influential contributions focus on enabling robots to robustly operate in complex, chaotic human environments. Deitke is best known for pioneering the Phone2Proc method, which uses a brief 10-minute smartphone scan to generate a realistic simulation of a user’s home, allowing agents to train and adapt before physical deployment—a breakthrough that directly tackles the simulation-to-reality transfer problem. This work, alongside his broader efforts on large-scale embodied benchmarks and datasets, has garnered significant attention, with his top-cited papers accumulating over 9 citations in a short span. Deitke’s research has been instrumental in advancing how embodied agents perceive, navigate, and interact with unstructured spaces, making him a key figure in the push toward truly generalist robots. His achievements include leading the development of open-source platforms that democratize embodied AI research, empowering students and researchers worldwide to build more capable, real-world-ready agents.