** Zoox is deploying sensor-equipped test vehicles as the first wave of autonomy in the capital. Credit: Zoox

While Elon Musk talks about a future in which he "drives everything himself," his subsidiary Zoox is quietly and methodically making this future the present. But they did not follow the simple path of testing in the quiet suburbs of California. Instead, they threw their symmetrical, rudderless and pedal—less robomobile into the very inferno - into the heart of US political life, Washington, DC. And this can be safely considered the most stress test for artificial intelligence after the citizenship exam.

Washington is not a city, it is a unique mix of turbulence. Here, protests are juxtaposed with tourist buses, diplomatic motorcades with ambulances, and pedestrians confident in their inviolability cross the road wherever they want, sipping lattes and looking at their phones. "Testing in a complex urban environment like Washington is invaluable,— said Zoox's CTO. "It contains all the possible scenarios imaginable: complex intersections, roundabouts, tunnels, bridges, and an extremely active and unpredictable audience."

What is Zoox and why does it look like a capsule from a science fiction movie?

Unlike the modified sedans of other companies, Zoox was designed from the very beginning as a robomobile, devoid of the concepts of "front" and "rear". It is symmetrical, with four wheels capable of turning in any direction, and two pairs of headlights. This allows him to maneuver in a confined space in a way that no ordinary car is capable of. Inside— there are two sofas facing each other, like in a subway car, creating an atmosphere for conversation rather than driving.

What exactly is being tested in Washington? Not driving, but thinking

The main task of the tests in the capital is not just to drive from point A to point B. It's about teaching AI to understand context and intentions.

Protocol and priorities: A robomobile should be able to recognize government motorcades and understand their special status, giving way to them the way an experienced professional driver would.

The sign language of the metropolis: Drivers in Washington actively use informal signals — flashing lights, hand gestures. The AI must interpret them correctly, distinguishing a friendly "drive through" from a warning "stop."

Pedestrian performance: Pedestrians here behave like the main characters. They may suddenly step onto the roadway, photographing the Capitol, or stop in the middle of a crossing, undecided. Algorithms should predict these trajectories not from a textbook, but from real behavior.

Logistics of the "last mile": One of the biggest challenges is where to stop for passengers to board and disembark in conditions of heavy traffic, forbidding signs and always busy taxi stands.

Security as paranoia: how to teach a robomobile

The company emphasizes that all tests at the first stage take place with two safety engineers on board — one in the driver's seat, the second in the passenger seat. They do not interfere in the process, but they are ready to take over at any moment. In addition, the car operates in a strictly limited geofence, which has been thoroughly studied and displayed in highly detailed 3D maps.

"We don't just teach the car the rules of the road. We're teaching him about urban culture," says one of the testing engineers. — The rules are just a skeleton. Muscles and nerves are an understanding of how these rules are violated and interpreted in real life. "

What does this mean for the future and where is the job here?

Successful testing in such a complex environment as Washington is a powerful signal for the entire industry. This proves that autonomous technologies are ripe for operation not in ideal, but in real conditions. And this opens the way for commercialization.

But behind each such robot car there is not only artificial intelligence, but also human intelligence. There is a whole army of specialists: machine learning engineers, data analysts, cartographers, cybersecurity specialists and, interestingly, "unmanned fleet operators" — people who will monitor the condition and routes of entire fleets of such machines. And it is precisely such high-tech vacancies that are already beginning to appear on specialized platforms, for example, on jobtorob.com where you can find not just a job, but a mission to integrate robots into our daily lives.

Meanwhile, the symmetrical Zoox capsules are quietly gliding through the streets of Washington, learning to live in harmony with the most unpredictable species on the planet — humans. And perhaps this experience will be the key to one day being able to get in a car, say "To the Capitol, please," and calmly immerse ourselves in our thoughts while the robot steers through the political and traffic chaos.