Google's new Gemini Robotics 2 platform allows for 'intelligent whole-body control'

Google DeepMind announced a new version of its artificial intelligence model Gemini that can control a range of robots including humanoids capable of tasks like cleaning up trash and picking up watering cans. The company released a video of the Gemini Robotics 2 platform in action showing robots putting a tape into a boombox, screwing in a lightbulb and tying up a garbage bag. The original Gemini Robotics system could already power bots that performed finicky actions like closing storage bags and folding origami but that was done via robotic arms and hands. This update brings "intelligent whole-body control" into the mix. Google promises the video features "real-time footage" of "fully autonomous" robots. The system uses several different AI models resulting in a multimodal understanding of surroundings including a vision language model and two vision language action models that control full-body and hand movement. Google DeepMind says it is another milestone in the path toward physical AGI. The model was specifically trained to perform every task in the video using a mix of human teleoperation, video examples and simulations. Google says it is taking a multi-layered approach to ensure future robots are safe with each model layer having guardrails and a new benchmark called ASIMOV-Agentic for detecting whether a command will lead to a harmful outcome. Carolina Parada, head of robotics at Google DeepMind, told Wired the safety question is even more pressing because the robots are put in many situations with uncertainty. Google is not about to roll out consumer-facing robots anytime soon.