🚗 מכוניות שנוסעות לבד: איך רכב 'רואה' כביש בלי נהג?

A car is parked on a street corner, there's no one sitting behind the wheel, and suddenly it moves — turns, stops at a red light, pulls into a parking spot. How does that work with no one holding the wheel? Self-driving cars "see" their surroundings by combining several kinds of sensors at once. Cameras spot road signs, traffic lights, and lane markings, much like human eyes. Distance sensors (like radar and lidar) send out invisible waves and measure how long they take to bounce back, letting the car "feel" exactly how close every other car, pedestrian, or lamppost around it is — even in darkness or fog, when an ordinary camera struggles. All of that information gets merged into a constantly updating 3D map of the surroundings, refreshed many times per second, which the car's computer uses to decide: slow down, turn, stop. It's important to understand that most "self-driving" cars on the road today don't actually drive completely on their own in every situation. In most of them, a human driver is still required to sit alert and ready, hands close to the wheel, prepared to take control at any moment if the system gets confused. Only a small number of companies currently operate truly driverless cars, and even then, only in limited, carefully mapped areas. The reason this is such a hard problem is that a real road is an unpredictable place. A kid might suddenly chase a ball into the street, rain can hide lane markings, another driver might do something strange or illegal. An experienced human can "guess" what other people on the road are about to do in a split second, based on a lifetime of experience. Teaching a computer to do that safely, in every kind of weather and every possible situation, is one of the hardest problems being worked on in AI today.