ChainThink report, April 14: According to 1M AI News monitoring, Google DeepMind has released Gemini Robotics-ER 1.6, positioned as a high-level reasoning model for robotics. This model demonstrates significant improvements in spatial reasoning and multi-perspective understanding compared to its predecessors ER 1.5 and Gemini 3.0 Flash, and is now available to developers via the Gemini API and Google AI Studio.

Core upgrades include three key capabilities: first, enhanced pointing accuracy enabling precise object detection, counting, spatial relationship reasoning, and motion trajectory planning, while correctly rejecting pointing to objects not present in the visual field; second, improved multi-view detection success, allowing the model to assess task completion by synthesizing inputs from multiple camera feeds, maintaining accuracy even under occlusion or dynamic conditions; third, newly added instrument reading capability, enabling interpretation of circular pressure gauges, vertical liquid level indicators, digital displays, and other industrial instruments through agentic vision for step-by-step reasoning.

This instrument reading capability stems from a collaboration between DeepMind and Boston Dynamics. On the same day, Boston Dynamics announced that Gemini and Gemini Robotics-ER 1.6 have been integrated into its Orbit AIVI-Learning product, which was rolled out to all AIVI-Learning customers on April 8. Following integration, the system now supports dashboard instrumentation, allowing the quadruped robot Spot to autonomously inspect industrial facilities and read pressure gauges and other instrumentation data.

Boston Dynamics stated that leveraging Gemini’s reasoning capabilities, AIVI-Learning has achieved improved baseline performance and accuracy across existing tasks such as visual inspection, pallet counting, and liquid accumulation detection. DeepMind emphasized that ER 1.6 is its "safest robotic model to date," demonstrating superior compliance with safety instructions in adversarial spatial reasoning tasks compared to ER 1.5; in safety risk identification tests based on real injury reports, ER-series models outperformed Gemini 3.0 Flash by 6% in text scenarios and 10% in video scenarios.