快猫短视频

Driverless cars could force other road users to drive more efficiently

Cars controlled by AI drive smoothly and more efficiently, and a computer model suggests the effect spreads to nearby drivers too
Self driving car
Self-driving cars may bring benefits to other drivers
Alex Kraus/Bloomberg via Getty Images

Autonomous cars are predicted to improve fuel efficiency for everyone on the road 鈥 an idea that will be put to the test on routes around Nashville, Tennessee, later this year.

at Rutgers University, New Jersey, and his colleagues previously used a computer model of a simple circular road with just one lane in each direction, and found that autonomous cars could decrease overall fuel consumption of all traffic by 40 per cent, even once adoption of these vehicles had only reached 5 per cent.

In subsequent real-world tests that mimicked this set-up, they proved not only that autonomous cars were able to drive more smoothly, but also that they forced human drivers to drive more smoothly. This meant there was less bunching of traffic caused by sharp braking affecting following vehicles, something that leads to phantom traffic jams and lowered fuel efficiency.

The team has now designed a larger computer model that investigates how driverless cars will affect more complex and realistic road networks. Piccoli says that the benefits are likely to be lower, but still present. He says that in a closed, circular system, a driverless car can make changes that propagate backwards through traffic and end up causing smoother traffic that benefits everyone 鈥 but on an open highway the cars will only be able to affect traffic behind them.

The team hopes to carry out a large-scale, real-world test with 100 driverless cars from various manufacturers in Nashville later this year to verify the results of the new model. The trial will have several phases, including one where the driverless cars act individually based on their programming, and one where they collaborate to ease traffic on a wider scale and share their own observations on traffic levels to a central server. 鈥淧eople will not even know they are in an experiment,鈥 says Piccoli.

The best-case scenarios from models 鈥渞arely happen鈥 in the real world, he says, but his team still hopes to reduce fuel consumption of all vehicles on the road during the trial 鈥 not just the driverless cars 鈥 by as much as 10 per cent.

鈥淭raffic doesn鈥檛 happen on a ring road and a single lane. It happens on open roads and multiple lanes. Reality is so complex and there鈥檚 so much nuance and unpredictability,鈥 says Piccoli. 鈥淚鈥檓 not gonna bet that we will reach that 10 per cent, but I hope that we will be able to show a really significant energy reduction. As you can imagine, if you take just the overall cost of the traffic system in any country, and you reduce that by even 5 per cent we are talking about billions of dollars.鈥

There could also be safety benefits, but Piccoli is careful not to make claims that aren鈥檛 backed up by data. In previous research, experiments revealed a reduction of more than 92 per cent in incidents of heavy braking when traffic was tempered by driverless cars.

at the University of the West of England says the concept has the potential to improve safety, fuel efficiency and travel times 鈥 but could also do harm if poorly implemented.

鈥淎 relatively large number of people will just follow the car in front,鈥 says Pipe. 鈥淎nd if that car in front is behind a car, which is behind a car, which is behind an autonomous vehicle that鈥檚 driving more smoothly, they may end up driving more smoothly themselves. But it鈥檚 a double-edged sword, because if it can improve something, there鈥檚 also potential for it to make things worse. It could generate lots of frustrated and irritated human drivers, who then act and do things that are not wise.鈥

Reference:

Topics: AI / driverless cars / Transport