Boston Dynamics Spot with Gemini Robotics

2026-04-14 · Publisher : Boston Dynamics · Domain : Robotics

Boston Dynamics plugs Google DeepMind's Gemini Robotics-ER 1.6 model into AIVI-Learning, Spot's visual inspection: reading gauges, displays and levels, 5S audits and leak detection.

### Background On April 14, 2026, Boston Dynamics announced that AIVI-Learning, Spot's AI visual inspection feature, now runs on Gemini Robotics-ER 1.6, a reasoning model released the same day by Google DeepMind. The robot no longer just photographs equipment: it interprets what it sees with a more general multimodal model able to reason across multiple views. ### Specifications - Safety and EHS checks: debris, spills, puddles, unauthorized people. - Asset monitoring: conveyor belt damage, sight glass levels, gauge and thermometer readings, more accurate digital display reading. - 5S audits and material tracking. - Immediate access for customers through Site Hub, a virtual machine or cloud hosting. ### Why it matters The deal illustrates the convergence of robot makers and AI model labs. Boston Dynamics, which had already teamed up again with Google for its Atlas humanoid in early 2026, hands the interpretation of Spot's images to a foundation model rather than to specialized models trained one by one. ### Key takeaways Spot becomes a concrete example of "physical AI" in production: a series robot whose inspection intelligence comes from a large general-purpose model.

Tags : Quadrupède, Inspection, IA, Google DeepMind, États-Unis