Google Connects Project Genie to Two Decades of Street View Imagery

*Google DeepMind is adding Street View data to Project Genie so subscribers can generate interactive simulations of real streets and environments.*

Google DeepMind has linked Project Genie to nearly 20 years of Street View imagery. The change lets users build simulated worlds that start from actual locations rather than abstract training data.

The update expands access to Google AI Ultra subscribers worldwide. It introduces controls for weather shifts, time of day, and uncommon events inside those simulated streets. DeepMind states the feature targets robotics research, game development, and virtual travel.

Prior versions of Project Genie created synthetic environments from scratch. The new connection anchors those environments to real-world geometry and textures collected by Street View cars over two decades. Users can now walk through or drive through familiar roads while altering conditions that would be difficult or unsafe to stage in physical tests.

Technical scope

The integration pulls from the existing Street View archive without requiring new data collection. Google describes the result as “new worlds anchored in reality,” where the underlying layout remains fixed but visual and environmental layers can be swapped.

No public benchmarks or accuracy metrics were released with the announcement. The company has not disclosed how much of the global Street View coverage is immediately available inside Genie or what resolution the simulations reach.

Access and rollout

Availability is limited to Google AI Ultra subscribers. The feature is described as rolling out globally following the May announcements from DeepMind and Google.

No third-party reactions or independent tests appear in the initial coverage.

The move gives robotics teams and game studios a ready supply of location-specific training scenes without the cost of custom scanning. It also raises the bar for what counts as a “synthetic” environment, since the base geometry now comes from documented public roads. Over time this could reduce the gap between simulation and deployment for navigation models, though results will depend on how well the added weather and scenario layers match real sensor data.

---

Sources:

• TechCrunch
• Google DeepMind
• Google

{
  "excerpt": "Google DeepMind adds nearly 20 years of Street View imagery to Project Genie, letting AI Ultra subscribers generate interactive simulations of real streets for robotics and gaming.",
  "suggestedSection": "ai",
  "suggestedTags": ["google-deepmind", "street-view", "world-models"],
  "imagePrompt": "An abstract urban grid of layered translucent streets and intersecting paths, with faint overlays suggesting shifting weather and time. Geometric forms and soft directional light imply simulation anchored in real locations, muted color palette, cinematic lighting, 16:9."
}