DeepMind announces Gemini Robotics ER 2, a robotics system designed to help robots understand video inputs, orchestrate tools, and perform coordinated work with multiple robots. According to DeepMind’s announcement and a separate recap, the system focuses on real-world robotic applications where robots must interpret visual information, decide how to use tools, and carry out tasks that may involve more than one robot.
Both sources describe Gemini Robotics ER 2 as a “step change” in three areas: video understanding, tool orchestration, and multi-robot collaboration. The coverage characterizes the system as enabling robots to reason about tasks from video, select and coordinate the use of tools needed to complete those tasks, and collaborate in multi-robot settings.
The articles do not provide additional implementation details or specific performance results in the supplied text. Overall, the shared message is that Gemini Robotics ER 2 extends the Gemini line of AI capabilities toward practical robotics, emphasizing perception from video and coordination for executing tasks in dynamic environments.
