Slashdot Mirror
Autonomous Road Train Project Completes First Public Road Test
theodp writes "Covered earlier on Slashdot, but lost in the buzz over the Google driverless car is Project Sartre (Safe Road Trains for the Environment), Europe's experiment with 'vehicle platooning,' which has successfully completed a 125 mile road test on a busy Spain motorway. Three Volvos drove themselves by automatically following a truck in the presence of other, normal road users. The Register reports that on-board cameras, radar and laser tracking allow each vehicle to monitor the one in front, and wirelessly streamed data from the lead vehicle tells each car when to accelerate, break and turn."
Well I hope they didn't send the "break" signal too often. That'd be a real bummer.
With mechanical linkages and a track instead of this complicated virtual 'pretend' train
That's ABSURD!!!!
General Relativity: Space-time tells matter where to go; Matter tells space-time what shape to be.
On long mountain roads far to often I see someone try to aggressively pass long sets of cars only to have to abort half way, causing other drivers to let them in quickly to avoid an accident..
I wonder if this road train would let them in.
EA David Gardner -"... but the consumers have proven that actually what they want is fun."
Seriously, how easy was that one? Brake is something that slows down a vehicle. Break is when it fails to Brake!
I can imagine sending the other vehicles the "signal" to head to my place for unloading. The security implications are tremendous.
Do you have ESP?
Who else read "Anonymous Road Train Project"?
Picturing guys in business suits with Guy Fawkes masks on.
I can't see the point in pursuing automated drivers. I mean, even if you could get them to work well 99% of the time, that 1% failure (or even .001% failure) would be just unacceptable.
I know we get computers to control aircraft, but it is a rather different situation. The problem of controlling an aeroplane with nothing up there to run into is a problem 10000 times easier than on the ground where there are so many hazards to avoid. The software would be so complex, there would be no way of knowing when it is going to plunge the vehicle into a tree. Odd happenings like this even occasionally happen to aircraft, but at least then the pilot usually has time to recover the situation before it is fatal. And that software is going to be MUCH simpler and auditable.
I worked on Carnegie Mellon's Red Team racing for a bit, but I didn't do anything major. I wanted to put in a redundant vision detection to their laser range finding and GPS guidance, but I got shot down. At least they let me poke around with GPS tweaking for a bit.
Anyway I always thought it'd be much easier to just make a train system where the rerouting sections get switched depending on your trip you programmed in. By being off normal rider roads, you'd only have to contend with other computerized trains, which could be tracked. The key thing at this point is just having some way to avoid deer and downed trees. I would think by first getting an automated train system up, then we could move into car systems later. The real trick is finding a city that doesn't have car transportation that wants to risk itself into automated trains. There are other problems with automated trains such as vandalism and terrorism and such.
God spoke to me
Just make sure we let the drivers know not to lay out on the hood of the truck, we managed to cover the story up by introducing the Google Driverless Car project, we don't want the general public to know..
We have had road trains here in OZ for many years http://en.wikipedia.org/wiki/Road_train Passing one is virtually impossible as the amount of air they drag along with them will either push you sideways or you get sucked under their wheels, I expect a close convoy of cars would have a similar drag effect.
I got to the chocolate box before you, that's why the hard ones have teeth marks.
As a pedestrian, longboarder and cyclist, I wonder how do those AIs fare with obstacles that are moving but are not cars.
I mean, what the fuck, human drivers are already pretty bad sometimes...
So the options are: (a) break the train --- but this is bad, because you're suddenly going to have to alert everybody from the break downstream that they're suddenly going to have to drive on manual, without much warning, or (b) maintain the train, but with a foreign car in the middle.
In either case, the cars in the train should identify the vehicle and notify the authorities. It would also help to update the traffic law to make it only legal to join a road train from the back with an approved autonomous tracking system. Anything else results in an expensive fine and a moving violation on the driver's record.
Convoy (1978): Truckers form a mile long "convoy" in support of a trucker's vendetta with an abusive sheriff. Based on the country song of same title by C.W. McCall. Trailer.
Those caravans of state highway escort trucks accompanying road work crews would only need one driver.
Most of the time highway traffic is safe and predictable. Driving 125 miles under favorable conditions (perfect weather and visibility if the news photo is any guide) without incident? Drunks do that and often get away with it; so do texting teenagers and fatigued truck drivers.
If someone demonstrated that he could drive 125 while smoking marijuana without having an accident, would we conclude that driving while high is safe and should be allowed?
The accident rate on highways is so low that 125 miles tells you nothing at all. The average accident rate in the United States is 8 fatalities per billion passenger miles. There is no way in the world a single 125 mile test involving four vehicles can tell you whether the accident rate for these car-trains is the same, ten times as high, or ten times as low. This is just a stunt, and proves nothing except that someone at Volvo had guts, and that someone in authority exercised bad judgement and allowed it.
"How to Do Nothing," kids activities, back in print!
On the road, hell is other people.
This is just a stunt, and proves nothing except that someone at Volvo had guts, and that someone in authority exercised bad judgement and allowed it.
TFA Said: The 125-mile test run was conducted at an average speed of just over 50mph and kept the three cars behind the truck at an average separation of 6m.
You might as well blindfold the drivers in the convoy, because at 50mph their reaction time works out to at least twice the 6 meter interval between cars.
Long story short, there is fuckall they could have done if something went wrong.
[Fuck Beta]
o0t!
How the hell is it both perfectly safe and also bad judgement for it to be allowed?
Of course it's a stunt, also known as PR.
Will this make them more exciting or less exciting. In the movies?
I recall back around 1990 the local (small town in Northern New South Wales, Australia) rubbish pick-up trucks had a master slave remote control configuration. The rear truck would just automatically follow the lead truck which had a driver. The trucks did not move quickly but they were master/slave with warning signs on the slave truck to warn other motorists placarded all over it.
So sure, maybe this new setup is a LOT more complicated, but its not all that new, preceded by some 20-odd years by our local council's rubbish trucks.
The accident rate on highways is so low that 125 miles tells you nothing at all.
I disagree. It tells you that the automated system will not instantly fail and cause the cars to drive into other cars/off the road/etc. Is that an acceptable substitute for long-term data involving thousands more miles and many more vehicles? No, but it will possibly allow for that data to now be collected, since they've demonstrated that the automated convoy isn't an instant death trap.
Is it a PR stunt? Obviously. But PR stunts can be useful. And if this particular one gets us closer to the sci-fi dream I've had ever since I was a kid of having a car automatically drive me to my destination while I read a book and relax in luxury, then I'm all for it.
This seems like a very cool idea, but I can't help but think that any system like this that goes "widespread" will be a prime target for crackers looking to do a little terrorism. Seeing as many public-use devices--pacemakers, and others that have been reported on Slashdot--utterly fail to be secure against such attacks--and since this one requires wireless receiving wireless signals in order to function, it's only a matter of time before an enterprising "cyberterrorist" decides to cause a pileup. Therefore, I question the wisdom of enacting such systems, even though the technology is cool, and could be very helpful if it worked properly (and reliably).
I often wonder if a Shweeb-like system, modified with partial motorization for long stretches of (high-speed) highway, uphill areas, etc. might be a more sensible approach. Sometimes advanced technology doesn't make things "better," despite how fun it is to think so.
Please tell me they have done a full set of failure mode tests before letting this shit lose on the public and isn't a case of 'the code compiles so ship it'.
Headlines label the muti-car accident the Lemming Effect, introducing yet another generation to the myth.
Given that the convoy depends on wireless communication, I wonder about its susceptibility to intentional jamming and hijacking.
Why is there no video?
you never drove a mile in spain, huh?
Nobody, not even Sartre, is claiming that this one test proves that automated driving is ready for prime-time.
A computer can perform those functions better than most humans, since it can track every single car nearby and their exact speed, trajectory, behavior patterns, etc
Especially if the other cars are designed to cooperate by broadcasting their speed, trajectory, etc. Then your car won't have to expend CPU cycles analyzing camera images to determine what the other cars are up to.
That that is is that that that that is not is not.