Home / News / Gadget News / Self-driving cars have been getting into fender-benders stateside

Self-driving cars have been getting into fender-benders stateside

Luke Johnson


Google Driverless Car

They’re supposed to make our roads safer, but new figures have revealed that a number of self-driving cars have been getting into prangs on American roads.

With just 48 autonomous vehicles currently being tested on Californian streets, four self-driving cars have been involved in crashes in the past eight months.

Three of these unfortunate incidents involved converted Lexus SUVs pimped out by Google, while the fourth belonged to parts supplier Delphi Automotive.

This might not be a sign of the machines’ uprising just yet, however. According to BusinessInsider, just two of the four incidents occurred when the self-driving cars had taken control.

The other occasions had a human at the controls.

Despite the multiple fender-benders, both Google and Delphi have denied their cars were at fault for any of the accidents.

In the incidents in which the driverless cars were running the show, the cars are said to have been travelling at speeds less than 10mph.

The Californian Department of Motor Vehicles has refused to comment on the cause of any of the accidents.

Related: Driverless Cars: Everything you need to know

Autonomous cars are currently banned from UK roads, although this could be set to change in the not too distant future.

Earlier this year, the UK government confirmed it would review the nation’s motoring legislation to accommodate driverless cars in summer 2017.


May 11, 2015, 3:33 pm

If everyone had driverless cars then there would be no problem. Driverless cars will do everything by the book, keep to speed limits, monitor conditions, accurately judge braking distances, etc. The problem is they can't combat human error and irrational behaviour 100%. There will always be situations where there is no clear escape or way to avoid an accident, in which case does the driverless car protect the passenger, pedestrians, or other road users?


May 11, 2015, 5:54 pm

Computers just lacks human intelligence. Regardless what is being said computers can't think like a human can that is why robots are still very basic. They take millions lines of code to do the simplest of things.

I doubt these driveless cars can monitor human expression. i.e. when other driver is giving way for you to turn right to relieve traffic on the opposite side.

There are so many examples I can give where driveless cars will have issues. For example during heavy traffic / rush hours at certain roundabout, unless you nudge out you'll be waiting forever causing huge traffic queue behind you. Some roundabouts you really have to boot your car through narrow opening gaps etc. A driveless car may detect the gaps to be short or will not nudge out because it thinks it's unsafe to do so. They will probably not boot the car either.

I just don't see driveless cars working. When robots can think for themselves and learn like humans do they will be far inferior in terms of intelligence. They have great computational power and does it extremely quickly but in terms of intelligence it is still the human that program or tell it what to do and how to react.


May 11, 2015, 6:53 pm

@Frankel, you contradict yourself.
First you say "lacks human intelligence", like that's a bad thing....
Then, you say "monitor human expression, i.e., when the other driver is giving way for you to turn". Expression is not the issue. I hope you (no more than I) would not take a chance on a turn across opposing traffic just because a driver raised his eyebrows, or pasted an obsequious smile on her face. If somebody is earnestly indicating that they are offering you their right-of-way, then they need to be making broad, and repeated, hand gestures. Even so, if the angle of their windshield (windscreen for you Brits) is wrong, all you will see is reflected sky, and no hint of what is going on in the driver seat of the opposing vehicle. Generally, it is a BAD idea to try to make yourself popular by relinquishing right-of-way. When you do that unexpectedly, you are doing it on behalf of all the drivers behind you, who might not agree, including the emergency vehicle that might be making its way up the shoulder, just as you encourage somebody to cut across.

The thing about driving efficiency and safety is that it is very often actions and choices that are counter to human tendencies (never mind whether it's jack-hole selfishness or tread-on-my-face "generosity") that produce the smoothest flow and the safest journey for everybody.

For example, in the paper today, a study confirms that "zipper" merge behavior (where everybody in both lanes remains in the respective lanes until merging at the very last moment, is both safer and more efficient than the other approach. Yet people seeing a merge in the distance will continue to merge early and feel righteous, while seething with anger at the "selfish" drivers who stay in the merging lane and get far ahead before merging just before the lanes join.

Machines can be (and are) programmed to maintain space commensurate with the capabilities of the vehicle, while also implementing non-intuitive, or non-"polite", behaviours that are more efficient for everyone.

A BIG bonus of having a proper robotic driver at the controls is that it doesn't get impatient and change its behavior (sorry, interface is changing my Canuck/Brit spelling to USian) on a whim or a temper tantrum. Yes, the robot will wait for a suitable gap. But it can be programmed to be flexible and to nudge in when the capabilities of the vehicle, the state of the road surface (is it wet? icy? painted?) permit it to get up to speed.

Think about what happens when you encounter a crowded roundabout. Your eyes dart back and forth, sizing up approaching gaps and vehicle speeds, while also keeping alert for pedestrians and cyclists and mopeds, and you are very likely to jump into a gap that strictly would not permit you to get in, if you could not count on other drivers to back off. It's judgment, but it's also an algorithm, and therefore can be programmed --- and tested and tweaked and tested again, and overridden when other factors intervene.
Also, what do robotic drivers do that more and more human drivers are beginning to think is also a good idea for us? Record, record, record. So, while you and I might have a purpose-built dash-cam, or a repurposed GoPro to record what idiots do in front of us, the robot keeps a record of the camera view out the front, the side, the other side, the back, and then the radar/sonar, infrared inputs, calculated speed and direction and acceleration vectors of all other nearby vehicles. Personally, if I saw a driverless vehicle (or one with a robo-driver logo), I would be VERY polite around it, since it would have a very comprehensive record of any misdeeds I might perform in its vicinity.

If I were the owner of the robo-vehicle/fleet, I might send such records to the police/traffic authorities as a matter of policy, or certainly to the automotive insurance industry association, who might very well be far more efficiently brutal to drivers/vehicle-owners who are recorded as behaving badly.

Actually, a version of that is already in the works, as most insurers now offer the plug-in tattle-tale that a policy-holder can use to have their driving behavior monitored in hopes of getting discounts on insurance premiums. As more and more people fall for this privacy intrusion, it will become the norm, such that people who don't allow ever-present monitoring will automatically be charged higher-than-baseline rates.

So, it's no big stretch that driverless vehicles will very shortly be a prime source of data on "the rest of us". Not long after that, we'll ALL be driving conservatively and efficiently, knowing that robo-nanny is always watching - if not from our own dashboard, then from that vehicle beside/behind/ahead of us.

Now, I could be wrong, but I'll be very surprised if it doesn't play out that way.

And with that cheery thought....

comments powered by Disqus