Until now, the capabilities of Google's Gemini AI have been limited to the digital realm. In order to be helpful for people in the physical world, AI is now set to be able to control robots on demand. — Photo: Andrej Sokolow/dpa
SAN JOSE: Google has unveiled two new versions of its AI system, Gemini, which are intended to lay the foundation for a new generation of robots that can perform helpful everyday tasks at home.
The first AI model is called Gemini Robotics. It enables the system not only to produce text, images or videos but also to control a robot’s actions.
The second model is named Gemini Robotics ER, where "ER" stands for "embodied reasoning". Google says this model enhances the spatial reasoning of Google Gemini.
Gemini Robotics-ER can develop entirely new abilities in an instant, according to a Google blog post. "For example, when shown a coffee mug, the model can intuit an appropriate two-finger grasp for picking it up by the handle and a safe trajectory for approaching it."
Google’s AI specialists at Google DeepMind are collaborating with the Texas-based robotics start-up Apptronik to build AI-powered humanoid robots that can operate efficiently and safely in dynamic environments.
Apptronik has already developed several robot models, including the humanoid robot Apollo, designed for tasks such as lifting, moving and stacking boxes in logistics and manufacturing.
Google says that for robots to be useful and helpful to humans, AI models for robotics require three essential qualities.
First, they must be all-purpose and able to adapt to various situations. Second, they need to be interactive to understand instructions or changes in their environment and respond quickly. Third, they must be skilful, capable of performing tasks typically done by human hands and fingers, such as carefully manipulating objects.
"Gemini Robotics represents a substantial step in performance on all three axes, getting us closer to truly general purpose robots," the company says.
Google also illustrated the capabilities of the new AI system in several videos. For instance, Gemini Robotics can sort objects into containers on command, even as they are moved across a room. Another example shows the robot folding an origami figure.
Google will also make its robot AI available to testers, including robot specialists Agile Robots, Agility Robotics, Boston Dynamics and Enchanted Tools.
The research project also covers the topic of safety. The physical safety of humans and objects near robots long been a fundamental concern in robotics.
This firstly focuses on avoiding collisions using traditional safety measures. Gemini Robotics-ER can be integrated with these safety-critical controls. Furthermore, the AI model can assess whether a potential action is safe in a specific context and trigger appropriate responses. – dpa
