Research / Embodied AI

Cross-Embodiment Learning

Why transferring skills between different robot bodies is a central problem in physical AI.

Why it matters

Different robots have different kinematics, sensors, degrees of freedom and end effectors. Cross-embodiment learning aims to make learned skills portable rather than retraining a model from scratch for every platform. Google DeepMind describes Gemini Robotics 2 as adaptable across robot embodiments, while NVIDIA describes GR00T as a cross-embodiment solution.

Primary source

Read the primary source →

Editorial research note · Reviewed September 2026