Gemini Robotics 2 Can Now Control an Entire Humanoid
Google DeepMind has announced Gemini Robotics 2, the intelligence layer it is putting behind the next generation of adaptable robots. The headline capability is whole-body control. Previous Gemini Robotics models drove only a humanoid's upper body for table-top tasks, and for the first time the model can now control an entire humanoid robot. DeepMind's own example is Apptronik's Apollo 2: asked to put a watering can into the green bin on the bottom shelf, Apollo walks to the table, picks it up, walks to the shelves, and places it.
Transcript
Google DeepMind announced Gemini Robotics 2 today, the first time one model can control an entire humanoid robot.
Earlier models controlled only a humanoid's upper body at a table. Now Apptronik's Apollo walks to a table, picks up a watering can, and places it on a shelf.
It controls a five fingered hand with twenty two degrees of freedom for delicate actions such as tying knots. Different robots can now communicate and work together on one task.
Google says movement speed and multi finger dexterity still need work. And this is not generally available. The reasoning model is in AI Studio. The rest is early access only.
Inboxsmith helps small businesses handle calls and messages so nothing gets missed. Please like and subscribe for more news.
Sources
Every claim in this video comes from the top ranking coverage of this topic. The claims and where each one came from:
- Gemini Robotics 2 is the intelligence layer powering the next generation of truly adaptable robots, unlocking intelligent whole-body control, advanced dexterity, and multi-robot collaboration.(Google DeepMind official announcement)
- For the first time, the model can control entire humanoid robots.(Google DeepMind official announcement)
- Previous models controlled the humanoid's upper-body to achieve table-top tasks.(Google DeepMind official announcement)
- Asked to put the watering can into the green bin in the bottom shelf, Apptronik's Apollo 2 walks to the table, picks up the watering can, takes a few steps to the shelves, and places it.(Google DeepMind official announcement)
- The model can control the five-fingered, 22 degree-of-freedom SharpaWave hand on the Apollo 2 robot to complete delicate actions like tying knots or sealing a ziplock bag.(Google DeepMind official announcement)
- Multi-robot collaboration enables different types of robots to communicate and work together to solve complex workflows a single robot could not do alone.(Google DeepMind official announcement)
- Google states its robots have more to advance in movement speed.(Google DeepMind official announcement)
- Multi-finger dexterous manipulation remains challenging, while whole-body and gripper-based dexterous tasks achieve a medium to high success rate.(Google DeepMind official announcement)
- Gemini Robotics ER 2, the reasoning model, is now available on Google AI Studio and in private preview on Gemini Enterprise Agent Platform.(Google DeepMind official announcement)
- The VLA and On-Device models are available to early-access partners.(Google DeepMind official announcement)