Gemini gains vision and real-world context via Volvo’s car cameras

Google’s Gemini is poised to see the world—not just through text or voice prompts, but through the lenses of Volvo’s upcoming EX60 SUV. At Google I/O 2025, the company announced a partnership that lets the AI-powered assistant tap into the vehicle’s external cameras, turning raw visual data into contextual insights for drivers.

A smarter assistant for parking puzzles and beyond

The first practical application may seem modest: asking Gemini to decode confusing parking signs. Instead of squinting at a symbol or recalling a vague road rule, drivers could simply query the assistant and receive a clear explanation. But Google’s vision extends further. With camera access, Gemini can analyze real-time surroundings—lane markings, pedestrian crossings, or even temporary road hazards—and provide actionable guidance.

Google emphasized that the integration relies on Volvo’s use of Android Automotive OS, Google’s embedded vehicle operating system. By embedding the assistant directly into the car’s software stack, Gemini gains a direct pipeline to the cameras without relying on a smartphone. This setup ensures faster response times and tighter integration with the vehicle’s other systems.

From signs to spatial reasoning

The EX60 isn’t just another SUV—it’s a rolling testbed for AI-driven mobility. Volvo’s decision to use Android Automotive positions the vehicle as a platform for real-world AI experiments. Google’s team hinted at future applications, including recalling specific road features, identifying available parking spots, or even assisting with complex maneuvers in tight urban spaces.

Early demonstrations showed Gemini processing a live camera feed to explain a no-parking sign’s restrictions. The assistant didn’t just read the sign; it cross-referenced the symbol with local regulations stored in Google’s knowledge base, then delivered a concise summary. This blend of computer vision and contextual data could become a blueprint for how AI assistants interact with the physical world.

Privacy and performance under the microscope

With cameras feeding data into an AI assistant, questions about privacy inevitably arise. Google stated that video feeds are processed locally on the vehicle’s onboard computer, minimizing cloud transmission. Only anonymized or summarized data would leave the car, and drivers retain full control over when the feature is active.

Performance is another critical factor. Google claims the system can interpret signs and surroundings in under two seconds, even in low-light conditions. The EX60’s hardware includes high-resolution external cameras and a dedicated AI accelerator to handle the workload without draining the battery.

What’s next for camera-enabled AI in cars

This isn’t Google’s first foray into camera-powered AI for vehicles. Earlier projects focused on dashcams or smartphone-based solutions, but the Volvo partnership represents a leap forward: a tightly integrated, always-on system that operates within the car’s ecosystem. As more automakers adopt Android Automotive, similar integrations could become widespread, turning AI assistants into true co-pilots.

For drivers, the benefits are clear: fewer parking tickets, less stress in unfamiliar areas, and a more intuitive way to understand the road. For the tech industry, it’s another step toward AI that doesn’t just answer questions—but sees, interprets, and acts in the real world.