Smart Robot Control New Google AI Gemini creates a new generation of helpful robots

Related Vendors

FAULHABER Drive Systems

Wuxi InfiMotion Technology Co., Ltd.

Google has now introduced two new variants of its AI system Gemini. These are intended to form the foundation for a new generation of so-called helpful robots.

The first AI model that Google has introduced is called Gemini Robotics. It ensures that the system can not only produce texts, images, or videos but also control the actions of a robot. The second model is designated ER, which stands for "embodied reasoning," as explained by Google. It expands the spatial “thinking” of Google Gemini. Gemini Robotics ER can, as emphasized, develop entirely new capabilities in an instant. For example, when shown a coffee cup, it can intuitively select an appropriate two-finger grip to lift the cup by the handle and calculate a movement sequence to approach it safely.

Three characteristics a general-purpose robot must have:

In the blog post from Google, it is stated that for AI models in robotics to be useful and helpful to humans, they need three essential characteristics! Firstly, they must be general in order to adapt to different situations. Secondly, they need to be interactive to understand instructions or changes in their surroundings and respond quickly. And thirdly, they must be skilled enough to carry out tasks that humans usually do with their hands and fingers—such as manipulating objects carefully. Therefore, Gemini Robotics represents a significant performance boost in all three areas and pushes forward the development of general-purpose robots.

The Texas-based robotics start-up Apptronik is involved in this effort

The AI specialists at Google DeepMind are working with the Texas robotics start-up Apptronik. The goal is to create AI-driven humanoid robots that can operate more efficiently and safely in dynamic environments. Apptronik has already developed several robot models for this purpose, including the humanoid robot Apollo, which is designed for tasks such as lifting, moving, and stacking boxes within logistics and manufacturing, as stated. In the previously quoted blog post, Google also demonstrates the capabilities of the new AI system in several videos. For example, Gemini Robotics can sort items into containers on command—even when these are being moved around the room. Another example shows the robot folding origami—a testament to its dexterity, corresponding to the third aspect. Google will now also make its robot AI available to testers, including robotics specialists Agile Robots, Agility Robots, Boston Dynamics, and Enchanted Tools.

AI-driven robots with built-in safety

According to Google, the research project also encompasses the topic of safety. The physical safety of robots and the humans in their surroundings has long been a fundamental concern for the acceptance of robotics. Initially, it involves using classic safety measures to avoid collisions. Additionally, Gemini Robotics ER can be connected to these safety-critical controls. With the AI model, it is possible to understand whether a potential action is safe in a specific context. If not, appropriate responses can be initiated.