Robots That Understand The World Are Coming – Google DeepMind’s Latest Model Is A Big Step Closer
This image was generated by AI and may not depict real events.
Google DeepMind has released Gemini Robotics-ER 1.6, a major improvement in spatial and physical reasoning for robots. The model enables robots to better understand their environment and make decisions based on visual input, with applications in facility inspection and other industrial use cases.
Google DeepMind has released Gemini Robotics-ER 1.6, a new model that improves spatial and physical reasoning for robots. Developed in collaboration with Boston Dynamics, the model acts as a high-level reasoning layer, processing visual input and producing instructions for lower-level systems. It shows significant gains in precise pointing, counting occluded objects, and multi-view reasoning. The model can also interpret analogue gauges and other industrial instruments, making it a concrete step toward robots operating in real-world environments. Safety reasoning has also been improved, with the model outperforming its predecessor on adversarial safety benchmarks. The release is a research and infrastructure development, available via Google AI Studio for startups building in the physical AI space.
This content was automatically generated and/or translated by AI. It may contain inaccuracies. Please refer to the original sources for verification.