Full Breakdown
Google DeepMind Launches Gemini Robotics 1.5: A Step Toward Intelligent Robotics
9/27/2025, 12:43:30 AM
Introduction to Gemini Robotics 1.5
On September 25, 2025, Google DeepMind unveiled the Gemini Robotics 1.5 series, comprising two advanced models: Gemini Robotics 1.5 and Gemini Robotics-ER 1.5. These models aim to enhance the autonomy and reasoning capabilities of robots, marking a significant advancement in the development of intelligent, general-purpose robots capable of performing complex, multi-step tasks in physical environments.
Core Functionality of the Models
Gemini Robotics 1.5 is a vision-language-action (VLA) model that translates visual information and instructions into motor commands for robots. It is designed to "think before acting," allowing it to break down tasks into manageable steps and explain its reasoning process. This model excels in executing tasks that require fine motor skills and contextual understanding, such as sorting objects based on specific guidelines.
In contrast, Gemini Robotics-ER 1.5 serves as an embodied reasoning model, functioning as the "high-level brain" of the robotic system. It is capable of planning, making logical decisions, and calling external digital tools, such as Google Search, to gather necessary information. This model has achieved state-of-the-art performance across various spatial understanding benchmarks, enabling it to create detailed multi-step plans for task execution.
Cross-Embodiment Learning
One of the notable features of the Gemini Robotics 1.5 series is its ability to facilitate cross-embodiment learning. This capability allows the models to transfer skills learned from one robot to another, regardless of their physical form. For example, a skill acquired by a bi-arm robot can be applied to a humanoid robot without requiring additional training. This innovation significantly accelerates the development of versatile robots that can adapt to diverse environments and tasks.
Safety and Security Measures
DeepMind has emphasized the importance of safety in the development of these advanced robotic models. Gemini Robotics 1.5 incorporates high-level semantic reasoning and low-level collision-avoidance systems to ensure safe interactions in physical spaces. The company has also updated its ASIMOV benchmark for evaluating semantic safety, with Gemini Robotics-ER 1.5 achieving top performance in internal tests.
Availability and Developer Access
Gemini Robotics-ER 1.5 is now available to developers through the Gemini API in Google AI Studio, while Gemini Robotics 1.5 is being offered to select partners. This rollout is expected to foster innovation in robotics, enabling developers to create more capable and intelligent robotic systems.
Criticism and Perspectives
While the launch has been framed as a milestone toward achieving artificial general intelligence (AGI) in physical robotics, some experts caution that the claims of "thinking" and "reasoning" may be overstated. Critics argue that the models primarily rely on advanced algorithms and data processing rather than genuine cognitive abilities. This skepticism highlights the ongoing debate about the nature of intelligence in artificial systems.
Conclusion: A New Era for Robotics
The introduction of Gemini Robotics 1.5 represents a significant step forward in the quest for intelligent robotics. By enhancing the reasoning and action capabilities of robots, Google DeepMind is paving the way for more sophisticated and adaptable machines that can perform complex tasks in real-world environments. As developers begin to explore the potential of these models, the future of robotics appears increasingly promising.
