Research

We are developing a new class of foundation models trained explicitly on safety constraints, enabling robots to reason about physical risks, geometry, and uncertainty in novel environments.