The Volkswagen Group is the first car manufacturer using a quantum computer to calculate traffic flows.
About 21 million inhabitants, more than five million cars – rush-hour in a megacity like Beijing can be hellish. On some days, hold ups last longer than four hours. This is why many residents call their city “Shoudu” – the capital of traffic jams – because the measures taken so far by the traffic authorities have brought little relief.
But how would it be if the congestion could be prevented before it was created?
This question was posed by Martin Hofmann, Chief Information Officer (CIO) of the Volkswagen Group, to his teams of IT specialists. With the help of movement data from 10,000 Beijing taxis, they have tried to calculate a sampletraffic simulation for a mega metropolis – “a highly complex calculation with billions of individual pieces of data,” as Hofmann emphasizes.
Traditional computers are overwhelmed
Data researcher Florian Neukart had to ask himself to what extent this is possible at all: “If there are too many factors in a given space, such as a large number of moving cars that you have to distribute to countless alternative points, this quickly leads to a combinatorial explosion, which overwhelms traditional computers, even with the cloud behind them,” he explains outlining the challenge. And since Neukart, in addition to his work as “Principal Data Scientist” at Volkswagen of America, also teaches quantum computers at the University of Leiden in the Netherlands, the possible solution to the dilemma was obvious: Could the combinatorial explosion be solved by using a completely different kind of computer which obeys the laws of quantum physics instead of those of electronics?
The computer that came in from the cold
In traditional computers, a bit can assume the value of either a zero or a one. A quantum bit – or qubit – can assume both at the same time, which exponentially increases the computing power of each qubit. In contrast to digital computers, the units of information in a quantum computer (bits or qubits) consist of tiny magnetic fields that are generated at temperatures close to absolute zero by very small coils in super-conducting circuits. Instead of magnetic fields, the energy states of individual atoms or the oscillation of light particles could be used, too. “It is possible to adjust an interaction between these qubits so that they can ‘sense’ each other,” explains Christoph Becher, a professor of experimental physics at Saarland University. “This system then naturally tries to arrange itself in such a way that it consumes as little energy as possible,” describes the quantum physicist. And extremely fast. Based on the distribution of the magnetic fields, attempts are then made to read the solution of the calculation problem. “The trick is to translate my question into calculation input that will enable the quantum computer to make the most of its advantage of speed.”
“In traditional computing you get the same answers to the same questions all the time“, agrees IT expert Bo Ewald, „but our quantum computer is probabilistic, like the universe.” Ewald, who was president of supercomputer manufacturer Cray Research and Silicon Graphics earlier, is now President of D-Wave International, the Canadian company that is selling the first commercially available quantum computers since 2011. “On the D-Wave system you don’t get the one answer, you get a distribution of answers, so that maybe 95 percent of them would mirror the digital result depending on the problem.” For some computing benchmarks, D-Wave is still slower than a common mainframe computer, says Ewald. But for specifically engineered problems it can be up to several thousand times faster than a super computer.
Optimize routes, control robots
Mathematical problems that quantum computers can solve better than digital computers include the optimization of routes, according to physicists such as Christoph Becher. This is why the Volkswagen Group is now the world’s first automotive company to use quantum computing. Since the beginning of the year, five Volkswagen software engineers from San Francisco and Munich have been programming on the D-Wave 1000Q via the cloud. “With the route optimization of the Beijing taxis, we want to show that we can solve a very practical problem using a quantum computer – and not just theoretically, as happens during research,” says Neukart.
Naturally, the goal is to use the insights for all customers. The taxis in Beijing merely represent a very suitable set of test data, he adds. "But quantum computers could also help us with other real-time critical applications," says Neukart, for example, in the lightning-fast self-optimization of robots in production, in autonomous driving, in networked manufacturing, in machine learning, and in intelligent mobility-solutions. The technology is still in its infancy, but is developing extremely quickly. Experts compare the state of development with that of electronic computing in the 1950s. For example, there are still no programming languages for quantum computers. “It is still hard to demonstrate that the system is doing what we want,” Neukart admits.
... and the traffic jams dissolve
It’s therefore no wonder that the optimization of the 10,000 taxis’ movement data did not work to start with. The numerous red dots, which each represented a taxi in the visualization, formed knots that were very difficult to unravel. It was only slowly, after the Volkswagen software engineers had split the city map into segments, that the red spots began to turn to yellow. “Our fifth solution was completely different from the first,” says Neukart, but it worked: Red turned to yellow, and yellow blobs became green. “The presentation is still simplified,” he admits. For quantum computing, however, it is still a big step forwards: “We have proven that it's possible!” The great possibilities that will result from this development in the future aren't even imaginable yet.
The Volkswagen Group at the CeBIT 2017