Should we take Google’s claim that it’s going to put self-driving cars on the market within five years seriously? Dutch automotive expert Henk Nijmeijer does not expect things to develop that quickly in the world of autonomous vehicles. He thinks cars should first learn to communicate with each other.
If we are to believe the latest reports, Google considers its self-driving car to be an ideal taxi – without a driver. If anything, this is a publicity stunt, assumes Nijmeijer, a professor of dynamics and control at Eindhoven University of Technology (TU/e) in the Netherlands. “Time and again they succeed in making the headlines with their plans, which they do very cleverly indeed. Google hopes that it can attract a major carmaker to do business with. And they will probably pull it off,” he says.
This does not imply that they are home and dry, though. “You should remember that the car can only do forty kilometres per hour and that no other cars are allowed within a 25-metre radius. You could call it a mobile library: it contains a database with various traffic situations and the appropriate responses to those situations. By means of a camera and radar it scans the situation and searches for it in its library. This takes time, even for the fastest systems. Hence the safety margin. That is far from practical, of course,” Nijmeijer explains.
The Google Car will perform fine when running its test rounds in the desert of Silicon Valley, but that’s very different from releasing an autonomous vehicle in the jungle of city traffic. There, Nijmeijer expects situations that cannot be found in the car’s library. “In that case, the car does not respond at all, which might well be the worst conceivable response,” he says.
Starting gradually with cooperative driving
The route towards autonomous vehicles on the public road will be gradual, Nijmeijer predicts. “You should begin with what we call cooperative driving. In this technique, vehicles have a wireless connection so that your car is told when the vehicle in front suddenly brakes or accelerates.” By means of a kind of sophisticated cruise control, the car reacts much faster to the speed changes of the car in front than a driver could do. This system is called Cooperative Adaptive Cruise Control (CACC).
“CACC means you don’t need legs anymore; you might say: you don’t need to step on the gas or on the brake yourself anymore,” Nijmeijer explains. The idea is that CACC can prevent so-called phantom traffic jams – which occur purely through poor anticipation.
In 2011, this method was tested extensively on the A270 between Eindhoven and Helmond. TU/e was closely involved in the organisation of the 2011 Grand Cooperative Driving Challenge (GCDC), also participating with a truck made available by the car manufacturer DAF.
“Next year there is going to be a new GCDC, in which we will be taking things a step further. The idea is that the steering will also be automated then,” says Nijmeijer.
Automatically following the vehicle in front
To this end, the participating cars will be fitted with cameras, which will be focused on the number plates of the cars ahead of them. This way, the car will ‘see’ when there is a curve in the road coming up and will automatically follow in the slipstream of the car ahead. At least, that’s the ultimate goal.
Nijmeijer admits that steering is more difficult than braking and accelerating. “If you turn your wheel abruptly when the car ahead does so, you may find yourself in the ditch, so to speak, which means you have to build in the appropriate delay. And merging is even more difficult, because there is no car ahead of you,” he says.
For the Grand Cooperative Driving Challenge 2016, TU/e has purchased two Renault Twizy vehicles. This small, dual-seat electric city car is entirely different from the DAF truck used in 2011, Nijmeijer agrees.
“Of course, it’s not a real car, but it’s electric and we consider that to be very important. The main thing is that it’s so straightforward. No power steering or ABS. This means you can simply build in an actuator to operate the steering wheel. And you are not hindered by existing control systems if you want to get going with the brakes,” he says.
Next up: designated driving on closed tracks
The next step towards ‘autonomous’ vehicles, the professor anticipates, will be specific applications on closed tracks. “Think of goods carriage in the port of Rotterdam, or passenger transport from the car park at Schiphol to the terminal.”
He also expects results to come out of an experiment whereby trucks can move in a column at night in the leftmost lane of the motorway from Rotterdam to the Ruhr area. Equipped with CACC, those diesel guzzlers could already save a considerable amount of fuel. And who knows, maybe the first car is going to be the only one that needs a driver – which would reduce the costs even more.
For his own Twizy fleet, Nijmeijer also has a dream destination. “I can see myself collecting visitors from the station with a column of Twizy’s, with me driving the one in front. The guests can then seat themselves in one of the empty cars, which automatically follow me to the TU/e campus or another location. For this you would actually need a bus lane, though. Genuine autonomous driving in ordinary traffic is no more than a speck on the horizon for now.”
Adapted from article by Tom Jeltes, TU/e Cursor
See also Google’s shopping spree.