Branton DeMoss

Papers

Total Citations

H-Index

About

Branton DeMoss is a robotics researcher whose work lies at the intersection of language-conditioned imitation learning and world models. His most notable contribution is the LUMOS framework, which enables robots to learn complex, long-horizon skills by practicing them in the latent space of a learned world model, then transferring those skills zero-shot to physical hardware. This approach bridges the gap between simulation-based training and real-world deployment, a critical challenge in modern robotics. With his most-cited paper already garnering attention, DeMoss is establishing himself as a rising voice in multi-task learning and skill transfer. His work demonstrates how language-conditioned policies can be combined with model-based reasoning to produce more adaptable and sample-efficient robotic systems. For students and researchers interested in embodied AI, DeMoss’s research offers a compelling vision of robots that can understand natural language commands and autonomously practice to master them—without ever needing to fail on a real robot first.