Boston Dynamics’ robot dog now reads gauges and thermometers with Google’s AI (arstechnica.com)

Boston Dynamics’ Spot robot can now read analog gauges and thermometers thanks to Google's DeepMind, which has developed the Gemini Robotics-ER 1.6 model for enhanced robotic reasoning. Announced on April 14, this new AI model enables robots to plan and execute tasks effectively, significantly improving their ability to conduct visual inspections and interpret complex instruments in industrial settings. With 98% accuracy in reading instruments, the Gemini Robotics-ER 1.6 marks a substantial upgrade from its predecessor, achieving a mere 23% accuracy, and underscores the potential of AI-driven advancements in robotics. The significance of this development lies in its potential applications across various industrial sectors, particularly in automating inspection processes. By integrating “agentic vision,” the model allows for sophisticated visual reasoning and interaction with physical objects, enabling robots to understand and manipulate their environments better. Additionally, with improvements in multi-view reasoning and image processing, the capabilities of robotic systems are expanding, paving the way for more efficient and versatile automation solutions in the industry. This evolution represents a crucial step forward in the intersection of AI and robotics, particularly for tasks that require intricate visual assessments in real-time.