Gemini Robotics-ER 1.6: Powering real-world robotics tasks through enhanced embodied reasoning

Google DeepMind's latest robotics model can now learn complex physical tasks from far fewer examples than before, a breakthrough that could finally make robots practical for real-world work beyond controlled laboratory settings. The system's enhanced reasoning abilities allow it to adapt to novel situations and execute multi-step instructions with surprising flexibility, suggesting that the long-promised era of general-purpose robots may be closer than many skeptics believed. With major tech companies racing to deploy embodied AI, the question is no longer whether robots will work—but how quickly companies can scale them into warehouses, hospitals, and homes.

Google DeepMind·May 5