Google has unveiled Gemini Robotics 2, an advanced artificial intelligence model aimed at empowering the next generation of humanoid robots. This innovative technology, developed in collaboration with DeepMind, is designed to transform how robots operate, equipping them with the skills to reason, navigate, and execute intricate tasks autonomously.

The introduction of Gemini Robotics 2 represents a significant shift from conventional robotic systems that are typically set for repetitive duties. According to Google, this model serves as an intelligence layer, enabling robots to interact more intuitively with humans and their environment while effectively managing multi-step tasks.

Google stated that Gemini Robotics ER 2 functions as a high-level brain for robots, facilitating natural conversations with humans and enhancing their understanding of the physical world. It prepares robots to break down complex tasks and assigns motor functions to lower-level vision-language-action (VLA) models.

Enhanced Control and Dexterity

One of the notable advancements in Gemini Robotics 2 is its ability to execute whole-body control. This enhancement allows robots to function more effectively, managing movements such as walking, crouching, and balancing all while executing hand movements simultaneously. Previous versions of the model were limited to controlling only the upper body of humanoids.

Gemini Robotics 2 has also seen improvements in dexterity. The model’s five-fingered robotic hands are now capable of performing intricate tasks, including tying knots and handling delicate items, while still providing support for simpler two-finger grippers used in warehouse environments.

By enabling a broader array of movements and skills, Google asserts that the new model can adapt to various robotic structures, allowing machines to operate effectively in dynamic settings without the necessity for extensive manual programming.

Project Management Functionality

The company has further developed Gemini Robotics ER 2 to act as a project management tool for robots. This feature enables robots to process spoken commands, segmented tasks into manageable parts, and monitor their progress. If complications arise, the model can adjust its strategy accordingly.

This intelligent adaptation permits robotic systems to undertake prolonged, multi-phase assignments, imbuing them with the capacity to make hundreds of decisions while collaborating with other machines in shared environments.

Additionally, Google released Gemini Robotics On-Device 2, enabling robots to operate independently of cloud support. This development permits robots to continue functioning offline and allows developers to tailor the model for new hardware with minimal training data.

Emphasis on Safety Standards

Safety considerations form a crucial aspect of Gemini Robotics 2’s development. Google has integrated both physical safety protocols and AI-based safety validations into the system. A newly established benchmark, ASIMOV-Agentic, has been implemented to assess whether robots can decline unsafe commands, signal for human assistance, and avert perilous situations.

Furthermore, the robots are equipped with proximity sensors, allowing them to detect nearby individuals, halt operations when necessary, and resume their tasks only when it is deemed safe. Google’s commitment to safety aims to ensure a secure interaction between robots and humans in various settings.

In summation, Gemini Robotics 2 is poised to revolutionise the robotics industry by enhancing capabilities, facilitating complex interactions, and ensuring safety for both users and machines.

Long or Short, get news the way you like. No ads. No redirections. Download Newspin and Stay Alert, The CSR Journal Mobile app, for fast, crisp, clean updates!

App Store –  https://apps.apple.com/in/app/newspin/id6746449540 

Google Play Store – https://play.google.com/store/apps/details?id=com.inventifweb.newspin&pcampaignid=web_share