Google Launches Gemini Robotics ER 2 for Autonomous Robot Control and Collaboration

Google has introduced Gemini Robotics ER 2, an artificial intelligence model engineered to serve as the cognitive system for autonomous robots. The model enables robots to perform real-time spatial reasoning, orchestrate complex multi-step tasks, and coordinate actions with other robots in shared environments. The system incorporates advanced video understanding capabilities, enabling robots to monitor task progress continuously and self-correct when issues arise. It demonstrates strong performance on task-critical metrics, achieving 57.4% accuracy in progress classification and 91.3% accuracy in identifying pivotal moments within video sequences, while maintaining the sub-second latency necessary for safe physical operations. Compared to its predecessor (ER 1.6), Gemini Robotics ER 2 delivers substantial improvements in execution speed and computational efficiency. The model is now accessible through the Gemini API, Google AI Studio, and private preview of the Gemini Enterprise Agent Platform. Google demonstrated the platform's capabilities through collaborations with Boston Dynamics, Apptronik, and Franka, showcasing multi-robot workflows and real-world task automation.