“BMW already sells cars that can follow the car in front, even braking, accelerating and steering by themselves”
How real or feasible is all this with today’s tech? Surprisingly close. Already, BMW sells cars that can – on a slow-moving urban road – follow the car in front, braking and accelerating and steering by themselves. They can do the same on motorways at high speed. Tesla’s Autopilot feature allows autonomous driving in a wider range of situations, but still can’t negotiate junctions or city streets. Mercedes has demonstrated that – using little more hardware than the sensors already fitted to its S-class – a car can drive through cities and rural roads. But to do so reliably its navigation system needs higher-resolution, fully updated data.
That next-gen mapping is underway now. Google’s autonomous driving fleet, extensively tested in several US states, points to the same thing. So most experts believe that autonomous driving will be technically feasible in a very few years from now. Sure, hurdles exist in the areas of traffic law and insurance (who’s responsible if an autonomously driven car does crash?) but they’re being tackled.
Some of the features of BMW’s concept – namely the warnings of hazards around the corner or hidden across a junction – rely on a level of external communication that’s as yet unavailable. Protocols are being developed across the industry so that cars will be able to distribute position and hazard data in spontaneous peer-to-peer networks. But standards adoption is very slow.
Over the years, what has happened instead is that cars’ ability to sense their surroundings has improved so fast that they haven’t needed as much externally supplied data as was expected. For example, many traffic engineers once proposed that every speed limit sign carry a transponder to tell passing cars the limit. Nowadays many quite mundane cars simply use cameras to read the signs themselves.