Research / Embodied AI

Embodied Reasoning

How AI systems reason about physical spaces, objects, goals and multi-step actions.

Why it matters

Embodied reasoning sits above low-level control: a system interprets a physical scene, decomposes a goal into steps, tracks progress and coordinates action with a robot policy. Google DeepMind describes Gemini Robotics ER 2 as a high-level reasoning model that plans multi-step tasks and can coordinate multiple robots.

Primary source

Read the primary source →

Editorial research note · Reviewed September 2026