Google’s Gemini Robotics Models Enable ‘Thinking’ Robots
AI
Robotics
Google
DeepMind
Gemini
Artificial Intelligence
Automation
8
Adaptive Robotics: A New Frontier
Media Hype
7/10
Real Impact
8/10
What is the Viqus Verdict?
We evaluate each news story based on its real impact versus its media hype to offer a clear and objective perspective.
AI Analysis:
While significant media attention is expected, the long-term impact of this adaptive robotics technology, particularly in enabling true autonomy and collaboration between robots, is substantial, though execution challenges will likely temper the immediate hype.
Article Summary
Google DeepMind’s latest advancements in AI robotics center around the Gemini Robotics 1.5 and Gemini Robotics-ER 1.5 models. These new AI systems allow robots, such as Apptronik’s Apollo and Franka Robotics’ bi-arm robot, to execute intricate, multi-stage tasks. Crucially, the robots can utilize web search to acquire real-time information, significantly expanding their capabilities beyond simple, isolated instructions. For example, a robot could sort laundry based on a local recycling guideline, or pack a suitcase according to the weather in London. The models are designed to ‘think multiple steps ahead’ and translate findings into natural language instructions. Furthermore, the technology allows robots to ‘learn’ from each other, transferring skills between different robotic configurations. This is enabled by Google DeepMind’s software engineer, Kanishka Rao, who notes that skills learned on one robot can be applied to another, regardless of their form. The models are being rolled out to developers via the Gemini API in Google AI Studio, representing a significant leap toward truly adaptable and intelligent robots.Key Points
- Google DeepMind’s Gemini Robotics 1.5 models enable robots to perform complex, multi-step tasks.
- Robots can now utilize web search for real-time information to aid in task completion.
- The new models facilitate robot-to-robot learning and skill transfer.