One can really only ever provide an interim report on autonomous driving. We still have a long way to go to get there. No, not another prediction, rather a kind of inventory. Drivers who drive on autopilot range from slightly annoyed to reasonably calm. But even the latter are anything but distracted , rather more attentive than self-drivers.

No, that's not just because the medium is new and people have to get used to it. No, it's to feel the tedious process, to explain the world to the autopilot. No, not just to explain, because unfortunately you often notice that it has not yet fully understood them.

Something is simply missing. Either e.g. all (!) streets are consistently and uniformly marked, then maybe an only camera-based system would have a good chance. But is that really a realistic demand? But probably not. And for so long, not only does the exclusively camera-based autopilot drive a little angularly, taking one curve a little too narrow for the taste of an experienced driver and the other perhaps too far.

That wouldn't be too bad if you didn't want to test such systems on two-lane roads. There the autopilot doesn't pay any attention to oncoming traffic in the curve, but always drives stoically the same way. This makes the heart of the driver or overseer beat faster, even if some people don't show it.

What we need is an accompanying system, perhaps derived from previous navigation. After all, the car is predestined to be able to go almost anywhere, to places where no external software will also not be available in the future. This is essential and ultrasound, radar and laser/lidar don't help there either.

The autopilot must know the curve beforehand, not have to 'experience' it first. Only if all test drivers with autopilot can sit completely relaxed behind the wheel for the entire journey, then level 3 of autonomous driving can be reached with some degree of certainty. Relaxed means that any danger is excluded, but of course not when there are blockades.

Imagine if the autopilot of an electric car would remain switched on in the city. It would find his way between the center line and the curb. But what happens when a car is parked at the curb as usual? There is a risk that it will stop behind it and wait for this car to drive off.

Neither your own sensors nor any navigation, no matter how precise, will help. This has to be continuously supplemented by sensor systems from transport partners. This is the only way the technology installed on traffic signs or traffic lights can tell the electric cars, watch out, there are no traffic lights and no stop signs. And the car is already on quite a lot of pictures of vehicles that came to meet it.

That would be a continuously updated navigation. In this case, their absence would not really be a tragedy, but just think of ice suddenly appearing on a bend or an accident. Ideally, construction workers have just set up the beacon or pylon and it is already being recorded by the networked system. And, of course, the sensors of the electric car that has just been receiving will also play their part when the occasion arises. Incidentally, this can also be the parked car itself.

Experts say that would already be harmless possible with the current bandwidth. How nice that we would have still another speed level if the would be underestimated. You can't really imagine the necessary size of the associated cloud, but you can already see from the current, still imperfect systems that it simply doesn't work without a standard that then connects everyone involved.

Yes, and then you could be amazed every day by the ability of artificial intelligence to evaluate millions of images of, for example, oncoming traffic, which concludes from their images that someone parks there. Because nobody is sitting in it, which of course you can't see from behind with integrated headrests. Only when AI and networking work to some extent can we start using autonomous driving carefree.