System identification
Infer compact, useful models of physical systems from indirect and imperfect measurements.
Physics-grounded intelligence
Models that learn from data
without forgetting reality.
Graybox Dynamics develops interpretable machine-learning methods for system identification, state estimation, simulation, and autonomous systems.
State estimation
Recover hidden physical states through model-based inference, sensor fusion, and uncertainty-aware estimation.
Simulation & sim-to-real
Combine physical structure and learned components to close the gap between simulated and deployed systems.