Self-driving systems are usually described in levels, from basic driver-assist features like adaptive cruise control, up to full autonomy where no human attention is needed at all. Most consumer vehicles on the road today sit in the middle of that scale — capable of handling highway driving but still expecting a human ready to take over.
A handful of companies now run limited driverless taxi services in specific, carefully mapped cities, where the vehicles have been trained on detailed local road data. Expanding that to unfamiliar roads, bad weather, and unpredictable human drivers and pedestrians is a much harder problem than it looks.
The remaining challenge isn't usually handling the ordinary drive — it's the rare, weird edge case: a couch that fell off a truck, a construction worker waving unconventional hand signals, a child's ball rolling into the street. Those low-frequency, high-stakes scenarios are what keep full autonomy a gradual rollout rather than an overnight switch.