In a major step forward for robotics, Google has introduced two new AI models—Gemini Robotics and Gemini Robotics-ER—designed specifically for robotic applications. Google announced in a blog post that these models, built on the Gemini 2.0 framework, aim to enhance robots’ ability to perceive, interact with, and navigate the physical world.
The Gemini Robotics model is an advanced vision-language-action (VLA) AI that enables robots to understand commands, interpret visual data, and execute physical tasks. This means robots equipped with Gemini Robotics can respond to spoken instructions, recognize objects, and interact more intuitively with their environment.
Taking this a step further, Gemini Robotics-ER incorporates spatial reasoning, allowing robots to analyze complex environments and adjust their movements accordingly. This model enables roboticists to leverage Gemini’s embodied reasoning abilities, giving robots a deeper understanding of their surroundings.
Key Features of Gemini Robotics and Gemini Robotics-ER
- Advanced Vision-Language-Action (VLA) Model: Enables robots to interpret visual data, understand commands, and execute physical actions.
- Spatial Reasoning Abilities: Helps robots navigate complex environments and adjust their movements accordingly.
- Embodied Reasoning: Allows roboticists to utilize AI-driven understanding for improved robotic interactions.
- Versatility: Can be applied across different types of robots, from humanoid robots to industrial automation systems.
A Versatile AI for a Variety of Robots
Google’s new AI models are designed for versatility, making them suitable for robots of various shapes and sizes—from humanoid robots to industrial machines used in factories and warehouses. The company has already tested the Gemini Robotics model on its bi-arm robotic platform, ALOHA 2, and has successfully integrated the model into Apptronik’s Apollo humanoid robot.
The introduction of these AI models is a significant development in the rapidly growing robotics industry. Over the past few years, advancements in AI and improved machine-learning models have accelerated robot commercialization, particularly in industrial settings.
Google’s Renewed Push into Robotics
Google’s involvement in robotics is not new. The company acquired Boston Dynamics in 2013, only to sell it to SoftBank in 2017. SoftBank later completed the sale of Boston Dynamics to Hyundai in 2021. However, the launch of Gemini Robotics and Gemini Robotics-ER signals a renewed focus on AI-driven intelligence rather than hardware development.
Google has also partnered with Apptronik to integrate these AI models into humanoid robots, aiming to create machines capable of both cognitive reasoning and physical dexterity. This partnership has been further strengthened by a $350 million funding round, led by B Capital and Capital Factory, with Google as a participating investor.
A Game-Changer for Robotics Startups
Beyond benefiting Google, these new AI models could significantly reduce development costs for robotics startups and speed up their time to market. AI-powered robotics models from industry leaders like Google and OpenAI allow cash-strapped startups to accelerate development and bring their products to market more efficiently.
