Here’s proof that autonomous car technology still has a long way to go. We nearly crash at 100km/h testing an automatic overtaking system.
Less than a week after an autonomous Google car crashed into a municipal bus in California, we got a frightening reminder why cars that can drive themselves are a long way from replacing humans.
I nearly put a brand-new $100,000 Mercedes E Class into a concrete barrier at 100km/h while testing the car's new automatic overtaking function in Europe this week, before it goes on sale in Australia later this year.
It was a stark reminder about the difference between “autonomous driving” and what the car industry is now trying to back-pedal and rename “driver assistance technology”.
The demonstration was supposed to be simple. Pull the cruise control stalk, indicate to change lanes, and then watch as the car magically steers itself gently into the next lane.
But as we learned the hard way, it does not work in all conditions. The car’s onboard cameras must be able to see line markings and, ideally, there will not be a sharp curve.
One of our first tests of the technology happened to be on a gentle curve on a crest. It should have worked. We were driving in perfect daylight and the road markings were clear.
Truly autonomous driving still has a very long way to go.
But it didn’t. Although the video shows the wheel moving slightly left and then slightly right, the steering was not moving quickly enough to avoid a crash into the barrier.
Fortunately, I grabbed the steering wheel just in time, and disaster was avoided. Mercedes got its car back in one piece.
For the record, we tested the technology several times before and afterwards and it worked perfectly.
Afterwards, we showed the chief engineer of this technology the video of our near miss and he said it was not clear if the assistance function was active (our video camera view did not show the symbol which indicates whether it was on or off).
I believe it was on, because the steering moved left and then slightly right before I grabbed the wheel. The engineer was not convinced.
Either way, it’s a warning to us all that “driver assistance” technology is not a party trick. And that truly autonomous driving still has a very long way to go.
In the US this week, Ford week filed a patent for an in-car movie screen to be used in an autonomous car of the future.
How about we wait for the industry to get the autonomous car right first? Right now, I’m not convinced.
The next time you read something about an autonomous car, just remember we are witnessing the gradual automation of the automobile.
Driving from Brisbane to Sydney to Melbourne and beyond while being able to read a newspaper or watch a movie is still decades away. If, indeed, it will ever happen.
Can you imagine trusting a car to steer itself for you? Tell us what you think in the comments below.