Study: Driverless Car Communication Improves Traffic Flow 35%

  • Nick Jaynes has worked for more than a decade in automotive media industry. In that time, he's done it all—from public relations for Chevrolet to new-car reviews for Mashable. Nick now lives in Portland, Oregon and spends his weekends traversing off-road trails in his 100 Series Toyota Land Cruiser.

Share on facebook
Share on twitter
Share on reddit

It’s widely accepted that autonomous cars will help alleviate traffic jams — or outright eliminate them. Able to communicate, self-driving cars will essentially see around corners and know what’s on the road ahead — a stalled car in the left lane, for example — long before onboard sensors can see it. And knowing what’s ahead and how to react will lessen roadway congestion.

  • Researchers at Cambridge University have built a live, but miniature, autonomous car testing facility to evaluate V2V communication systems.
  • A recent study found vehicle-to-vehicle communication improves normal traffic flow by 35% in cases of stalled vehicles occupying a traffic lane. 
  • It also found a marked improvement in traffic when self-driving cars yielded to aggressively driven human-driven vehicles.
  • The second finding raises the question, will a few aggressive humans slowdown self-driving enough, to hasten the demise of non-autonomous cars? 

Although the concept has been regarded as true, it’s not really been demonstrated yet. That’s because autonomous car fleets aren’t yet large enough to physically put into practice to test the theory. Therefore, no one was exactly sure how much smoother traffic will flow once self-driving cars dominate roads. That is, until now.

Researchers at Cambridge University revealed this week that during a small-scale experiment — and I mean small — they were able to demonstrate a 35% improvement in traffic flow. That is, when the vehicles were cooperating. During a scenario when a vehicle was driving aggressively, disrupting normal flow, vehicle-to-vehicle (V2V) communication enhanced traffic movement by 45%.

Student scientists fitted scale models of production vehicles, which appear to be Range Rover Sport models, with motion capture sensors. They also installed a small computer called a Raspberry Pi on each car, which enabled them to communicate with one another over a Wi-Fi signal. Running algorithms created by the study co-authors, the cars reacted to different driving scenarios — from cooperative to ‘egocentric’ aggressive driving.

As you can see in the video above, when a car stops unexpectedly in in the lane, with V2V disabled, a traffic jam quickly builds up. Once the cars are communicating, however, they are able to know what’s ahead on the road and plan and react preemptively, saving time and space.

The same goes for when one of the researchers begins to drive one of the cars remotely in an aggressive fashion. Knowing a virtual maniac is zigzagging on the road around them, the cars can give it the right of way and avoid would-be traffic snarls.

First off, I think this might be the cutest autonomous car demonstration I’ve ever seen. I absolutely love these little models cruising around on the carpet.

Putting just how adorable the experiment is aside for a second, however, it highlights a couple points about the future of autonomous driving that we ought to discuss. What’s more, it highlights an implication that perhaps the study co-authors glossed over.

Undoubtedly, this study demonstrates that roadways will be safer when V2V communication is paired with automated driving. Hopefully it will also inspire not just universities but also automakers to test and refine their self-driving tech on a smaller scale.

That said, I do see one issue in the findings. And that falls in the egocentric, aggressive driving demonstration. To me, it shows that self-driving cars could be incredibly annoying to ride in. That’s because manually driven cars could essentially dominate automated vehicles on the roadway.

Think about it. We’ve all seen the car ripping through traffic, jumping lanes, speeding, and generally acting like a jerk. A lot of us ignore them. If autonomous cars know such a driver is coming, though, and they give the right of way, self-driving cars are essentially letting the jerks win. An egocentric driver could really just take the road for themselves, knowing that autonomous cars — erring toward safety — will let the human driver essentially ride roughshod over the robo-car.

I’ll freely admit I am being a bit hyperbolic. It’s a good example, though, to remind us that autonomous cars will be, at least at their outset, the most timid drivers on the road. Unless you’re ready to ignore the road altogether and just enjoy the slow cruise, a self-driving car might be more irksome for type-A drivers than it’s worth.


About the Author

  • Nick Jaynes has worked for more than a decade in automotive media industry. In that time, he's done it all—from public relations for Chevrolet to new-car reviews for Mashable. Nick now lives in Portland, Oregon and spends his weekends traversing off-road trails in his 100 Series Toyota Land Cruiser.

Close Menu