tech
February 5, 2026
Gemini Robotics
Understands the physical world, and adapts and generalizes its behaviour to fit new situations. Breaks down goals into manageable steps to make longer-term plans and overcome unexpected problems.
TL;DR
- Gemini Robotics 1.5 is a vision-language-action (VLA) model that translates visual information and instructions into motor commands for task execution.
- Gemini Robotics-ER 1.5 is an embodied reasoning model specializing in understanding physical spaces, planning, and decision-making within its environment.
- The system supports a dual-model approach, integrating VLA and ER models for comprehensive robotic capabilities.
- Gemini Robotics enables robots to perceive, reason, use tools, and interact with humans, solving complex tasks autonomously.
- Key capabilities include generality, agentic behavior (tool use, planning), thinking before acting, interactivity, dexterity, and adaptability to multiple robot embodiments.
- Responsible development is a focus, with practical safeguards and collaborations with experts, policymakers, and a Responsibility and Safety Council.
- Partnerships include Apptronik for humanoid robots and collaborations with over 60 trusted testers.
- Gemini Robotics On-Device and the Gemini Robotics SDK allow for local optimization and adaptation by developers.
Continue reading
the original article