Gemini Robotics ER 2: Google DeepMind научила роботов рассуждать и работать в команде

Google DeepMind has introduced Gemini Robotics ER 2 — a next-generation model for controlling robots in the real world. Key capabilities: video stream understanding, orchestration of multi-step tasks, and coordinated operation of multiple robots simultaneously.

What Gemini Robotics ER 2 Can Do

Gemini Robotics ER 2 is built around three directions, each of which Google DeepMind calls a fundamental step forward for robotics applications.

Video understanding — the model analyzes visual context through a standard video stream in real time: it recognizes objects, tracks motion, and makes decisions on the fly. The robot no longer needs a rigid pre-defined map of its environment or a set of specialized sensors.

Task orchestration — the system plans multi-step scenarios, using multiple tools in the right order. This is a transition from a single command to goal-oriented planning: the robot sees the task as a whole and builds the sequence of steps on its own.

Robot collaboration — multiple machines act jointly: they distribute subtasks, coordinate actions, and adapt to changes in real time. According to DeepMind, this is precisely what was missing to move from controlled demos to real production and logistics scenarios.

• 3 key directions: video understanding, task orchestration, multi-robot collaboration
• Developer: Google DeepMind
• Positioning: real-world tasks, not just lab tests
• Lineup: development of the previous Gemini Robotics ER generation

Why Robot Collaboration Is an Unsolved Problem

Traditional industrial robots work in isolation: each performs a strictly defined operation according to a rigid algorithm. Coordinating multiple machines in real time requires more — a shared understanding of context, the ability to redistribute tasks when conditions change, and awareness of a partner's actions.

This is exactly where AI models change the equation. Instead of manually written scripts, they allow robots to reason about a task and adapt to changes on the fly — without reprogramming for each new case.

"Gemini Robotics ER 2 helps robots reason, interact, and solve real-world problems.

This is a step forward in video understanding, tool orchestration, and multi-robot collaboration," states the official Google DeepMind blog.

According to the company, the model targets real-world applications — not demonstrations in a controlled environment. This is a significant claim: most robotics systems today still fail to achieve the necessary reliability outside the laboratory.

ER 2's Place in the Race for Physical AI

Google DeepMind is building Gemini Robotics as a lineup covering the entire robot control stack — from perception to planning and cooperation. Gemini Robotics ER 2 simultaneously expands capabilities across all three fronts, which distinguishes it from point updates.

Physical AI is one of the fastest-growing areas of competition among leading laboratories. The emphasis is shifting from specialized industrial machines to universal systems capable of adapting to new tasks without rigid reprogramming for each scenario.

What This Means

Gemini Robotics ER 2 is a signal that Google DeepMind is moving from language and visual AI systems toward physical intelligence: robots capable of reasoning, coordinating, and working in the real world rather than in a laboratory. If the declared capabilities are confirmed in practice, this will significantly expand the applicability of robots in manufacturing, logistics, and related industries.