Google DeepMind has introduced a new generation of artificial intelligence under the name Gemini Robotics 2. The suite of models is designed to make robots, and humanoid designs in particular, considerably more independent. The focus lies on controlling the entire body, performing more refined movements and enabling several robots to work together.
Demonstrated on Apollo 2
The technology is being shown on the Apollo 2, the humanoid robot built by Apptronik. According to the developers, the robot can now carry out autonomous full-body movements, including walking, crouching and bending. The system coordinates arms, legs and torso together rather than driving individual parts in isolation.
This integrated approach marks a clear shift from earlier methods. Where robots were often programmed joint by joint or task by task, Gemini Robotics 2 treats movement as a coherent whole. That makes for smoother and more stable actions.
What it means for manufacturing
These developments matter for production environments. Humanoid robots that can move independently and handle delicate tasks may be deployed in workplaces designed for people, without major changes to the surrounding infrastructure. The emphasis on dexterity opens up possibilities for jobs such as assembly, material handling and quality control.
The option for multiple robots to collaborate also points to scenarios in which machines tackle more complex assignments together. That could increase the flexibility of automated production lines.
- Whole-body control for more stable movement
- Improved dexterity for delicate operations
- Collaboration between multiple robots
How quickly these capabilities reach everyday factory use remains to be seen. Still, the combination of advanced AI models and a physical humanoid robot illustrates the direction the sector is taking: toward robots that depend less on fixed programming and can respond more independently to their surroundings.