Automated Cars

[npts2020]

Every year America's highways kill 10 times the number of people who died on 9/11, and in every bit as gruesome a manner. The overwhelming cause of this carnage is the fact that a human operator is easily distracted and unpredictable. The obvious solution would be to take the human driver out of the equation and automate the system. My question is what would be the best system of automation? Two possible solutions are to use tracks (like rail, monorail, or overhead systems), which are expensive to build, or imbed an optical or magnetic grid in the road (or beside or above) for onboard sensors to follow and translate for directional purposes. I would appreciate any feedback about the practicality of either system and ideas for other types of control.

[iNow]

Another benefit would be how much faster we could all travel... Like a super fast elevator turned sideways, we'd all be zipping along at proper intervals and no need for these complex coordinated symphonies of various human intentions, emotional states, and driving abilities all on one "stage."

 

I think we can do this with GPS technology. Much like the chic on the Garmin knows when to tell you to turn left or right, the car could use that same data to know when and where you are and what's going on ahead. No magnets or rails needed...

 

One issue is foreign debris on the road, like a fallen tree or a dead animal or the like... Not sure how to handle that, but I do think the technology behind GPS would be the best starting place.

 

I actually think companies like Volvo have already been testing ideas like this, but I'm not really sure.

[insane_alien]

well you can use computer vision systems to detect foreign objects but those are still very computationally expensive as well as financially expensive. cars would also have to 'talk' to each other as well so that each of them knows what the other is doing(for the same reason we have all those indicator and brake lights).

 

we are approaching the stage where this is technologically possible but i think it will be quite some time before it is economically viable. also, there will be a bunch of people(me included) who will want to retain manual control of their car.

[Mr Skeptic]

The system would have to detect thinks like a little kid running into the highway to fetch his ball, or a deer. It would have to detect them before they were on the highway, as they could run in front of the car and there would be little time to dodge. I'd agree that we can take care of navigation via GPS by now. Construction zones might be another problem.

[big314mp]

The system would have to detect thinks like a little kid running into the highway to fetch his ball, or a deer. It would have to detect them before they were on the highway, as they could run in front of the car and there would be little time to dodge. I'd agree that we can take care of navigation via GPS by now. Construction zones might be another problem.

 

What if you prevent them from running onto the highway in the first place? I'm thinking a security fence alongside or something.

 

Or maybe we just trust that people won't run across highways. Sort of how we deal with railroad tracks, and, well, highways.

 

For construction zones, perhaps an automated detour?

[iNow]

AFAIK, most GPS navigation systems already adjust for construction zones.

To add to and supplement the GPS, lane sensors might need to be placed in the roadway, though.

[big314mp]

I'd feel uncomfortable with a GPS only system, considering that slight inaccuracies (18 inches, even) could kill people. And what happens if the signal is dropped?

 

I'm thinking a road based system to guide the car, and GPS to tell it where to go.

[insane_alien]

big314mp, it would never be a fully gps system and not only that, but if every care depended on the GPS system for navigation it would be made far far more accurate.

[npts2020]

The system I would like to see designed should be able to exceed 300 kph and therefore ought to be enclosed (one of the main things that would also make it safer). I would think efficiency, reliability, and safety would best be served by having a large main computer telling on board computers what to do (any other opinions about this?). Onboard computers would only control car functions, taking all speed and direction orders from the main. Ithaca, New York is set to start planning for a similar system to the one in Morgantown, W. Va. but IMO they have set their sights too low and should have smaller faster cars. I like the idea of sensors imbedded in the road for flexibility and cost but don't really know enough about them to know how practical they would be at higher speeds. Rails are proven to be safe and effective but are also expensive and more limiting. GPS would likely be used as a status check more than control unless some way of being in constant contact and control of millions of vehicles is constructed and accuracy further improved. I know little about the hardware for GPS but could the current system even handle that much information traffic?

Back in the mid-1970's the U.S. DoT did a study about automating the highways and concluded that the main technological (and almost only real) obstacle was a control system. Fast forwrd 30 years where many hand held devices have more computational power than the best computers of the time of that study and you have overcome the main hurdle for effective controlling tools. The only thing that has to be done today is apply that computational power to the required use. That is the current question, what is the best control system in terms of cost, reliability, compatibility with retrofitting current vehicles etc.? I am as interested in your reasoning as your opinions.

[Psycho]

The system would have to detect thinks like a little kid running into the highway to fetch his ball, or a deer. It would have to detect them before they were on the highway, as they could run in front of the car and there would be little time to dodge.

Well the whole system would mean you could have the other cars around you swerve as well to make room, which would never happen now you would just hit the other cars.

 

To be fair if you couldn't see the kid coming a computer with sensors could break much fast due to a lesser reaction time.

[big314mp]

I think kids running across highways is a virtual non-issue, that can be dealt with the same way highways keep kids off of roads: common sense and barriers.

 

Now, what happens if/when a car breaks down on the road? You would have to have some way to rapidly clear a broken down vehicle off of the road.

 

Or perhaps you could do something like what (I think) npts is advocating, except that both control and power come from the external track. Rails are the most viable option for that.

[khaali91]

we'd also have to remove the several billion cars through out the world!!

[iNow]

we'd also have to remove the several billion cars through out the world!!

 

Not necessarily. The same end could be achieved by removing their source of fuel. If the petroleum goes away, the cars become a very large paper weight.

[npts2020]

we'd also have to remove the several billion cars through out the world!!

 

How many of those several billion cars are going to still be operating in another 20 years anyway? The whole point is to make the system safer, faster, more efficient and convenient and to replace those cars with a more sustainable means of transportation. But that is a whole other discussion. Any system I can envision will need more than one lane of traffic to handle peak volumes even with automation. Rails are very good for close tolerances and efficiency (i.e. rolling friction) and might have better fail-safe capabilities but they are very inflexible and expensive. For those last two reasons it would be better to have some sort of optical or electromagnetic guide for control if the fail-safe and close tolerance problems can be brought to a level similar to rail.

[CaptainPanic]

It's one thing to replace all cars (they're made of steel, and that's fairly easy to recycle). To replace the roads as well would be a different thing altogether.

 

You can replace the highways and popular secondary roads with rail systems, or place overhead whatevers. But it's not economical to replace even the dirt roads and forest paths... and think about the complexity of a large parking area? A large slab of asphalt or concrete is just so much cheaper.

 

To reduce the fatalities on the road, I think that some form of object detection is needed. These are being developed at the moment. It just detects any object that has the potential to hit the car (that can include a deer running onto the road). These systems are just that 0.5 second faster than the human driver. But you cannot reduce the risk to zero simply with such a system. And a fence along every road is also no option.

 

Replacing the driver completely with any GPS or other navigation system is overdoing it. How many fatalities are caused by people who are lost? Compare that to the number of computer bugs you've seen in your life. I'd rather navigate (steer the car) myself

 

Concluding: I don't see the point of replacing the driver completely. But drivers can use a little help in dodging objects that suddenly appear on the road.

[big314mp]

If you use GPS (and actually put some real effort into bug elimination) you can use that to guide where the vehicle goes. Use radar to measure how far the car in front is, to adjust following speed. If you bury two wires in the road, and build a car that can detect where those wires are, then you can have a car that stays in a lane. You can also do that without completely replacing all of the roads. You could probably do it when the roads are being resurfaced, and thereby save on tearing up the asphalt.

 

You really only need to do this on major roads an highways, as backroads are probably better off under manual control.

 

As for the fences, the most of the highways here have fences and/or sound barriers already. It's not a huge issue to install them.

 

Which raises another issue, IMO. What about noise levels of traveling so fast?

[YT2095]

Which raises another issue, IMO. What about noise levels of traveling so fast?

 

I tend to find that the noise is Greater at lower speeds than higher ones, both in the cage and out of it.

of course this does all depend on the road surface and weather conditions (and probably a few Other external factors as well) etc...

although this is a personal opinion only, some other ears may differ?

[big314mp]

Well, on the highways here, there are sound barriers built along the sides of virtually all of the highways. At 300km/h, road noise and wind noise are going to be serious problems barring new technological developments.

 

I have heard of a new type of asphalt that helps reduce road noise, so the problem isn't necessarily unsolvable.

[CaptainPanic]

I'd say that the fuel consumption will be the greater problem at such speeds. We're talking about cars, not trains... they each have their own front end, which means they each will be attempting to break through the air (greater air resistance than a single train carrying 500 people).

 

I think the best argument for automated systems is not accident prevention. It is that cars can drive closer to each other (possibly because they can communicate with each other). This can greatly increase the capacity of a road... and in turn greatly decrease traffic jams.

[big314mp]

If you get the cars to drive close enough to one another, you can have them slipstream one another to increase fuel efficiency.

[iNow]

If we spend the investment of time and money making our automotive system so automated that human inputs are not required, do you all really truly believe that we would still be running our cars on petroleum by the time such a system was all in place?

 

Think batteries and whatnot. That technology is coming more quickly than you know, and I have a very hard time believing that we'd still be burning dino fuels to make our autos move if we were simultaneously willing to make the enormous investment toward automation.

 

Basically, if we were going to automate cars and roadways, I have a sneaking suspicion we'd also be improving the power source of our cars in parallel (which is already happening independent of the automated roadway idea anyway).

[npts2020]

No options for improving our highway system will be cheap. The sentiment about not wanting to turn over control for automation is fairly common as well. However, as anyone who has driven around a major American city during rush hour can attest, the highways are basically bursting at the seams. The biggest problem isn't that there aren't enough roads, it's that the roads are inefficiently used in large part because of the extra safety margins required by human drivers. It is not possible to automate every highway at the same time, either, but once begun should be done as quickly as possible. I would be happy to discuss the politics and economics of such a project on another thread if anyone is so interested but I would like to try to restrict this discussion to the technical aspects. The reason the thread was started in reference to control systems is that this is the main untested part of automated traffic control. If I had unlimited resources I would actually test as many control regimens as could be thought of but it seems that just building a scale model is going to be no easy task for what I have access to at present. The reasons I think such a system is both necessary and feasible at this time mostly have to do with safety and environmental concerns and the fact that the control system that would be required no longer needs new technology. A bonus for an automated system is that you can power the system from any source, affording the opportunity to substantially convert about a quarter of our national energy use to renewable sources. If the system is to be high speed it will need to be enclosed, current interstate highways are approaching the limit IMO of what an open system can safely achieve. Enclosing most of the system will enable higher speeds by not having to react to obstacles or being affected by weather. Solar panels and wind mills enough to completely run the system can be erected on top and a new utility grid on either side, this should reduce noise as would better aerodynamics for vehicles using the system. It should be possible to use retrofitted vehicles as well with an optic or electromagnetic, guide system making such more versitile, but those vehicles would never likely use the fastest lanes of the system, only freight and merge lanes. My leanings are toward imbedding magnets/wire/optical sensor in the pavement and/or ceiling and/or walls over guidance by rails but nearly everything I read about "advanced" transportation talks about rail/light rail.

[Mr Skeptic]

If you get the cars to drive close enough to one another, you can have them slipstream one another to increase fuel efficiency.

 

Or you could have someone cause a big pileup accident/sabotage.

[npts2020]

Or you could have someone cause a big pileup accident/sabotage.

 

On an automated system, such as I have described, an event like this should be exceptionally rare and not as bad as many of the pileups that frequently occur during bad weather and some other times on the current system. For one thing a computer will automatically make all vehicles that might be affected react uniformly and faster than any human. Furthermore, a computer will make the "right" (whatever was programmed to be correct anyway) decision every time and without hesitancy. The situation you describe happens many times every year on our current highways, i would think once a year on an automated system to be an unacceptably high rate.

[waitforufo]

The IEEE has a Vehicular Technology Society which publishes a peer reviewed journal on this topic. (This journal also covers wireless technology back from the days when wireless phones were called "mobile phones.")

 

Recent trends in this area include vehicle positioning and wireless vehicle-to-vehicle communication. Positioning is done by GPS, Wireless (vehicle location relative to base stations), and location RF tags. Location RF tags seem to be the hot item at the moment. At one time people considered imbedding these in the roadway lane centers, but today they talk more about positioning them periodically or pseudo randomly along the roadway. This simplifies servicing. Wireless vehicle-to-vehicle communication would be done with something like Bluetooth or Zigbee (but faster) creating an ad hoc mesh network amongst vehicle neighbors. That way if someone many vehicles ahead of you does something, your vehicle knows about it. Control systems are then built to maintain speed, keep safe distances between vehicles, and keep everyone on the road.

 

Problems? Well, if someone doesn't have the equipment or their system is broken, they are not part of the mesh. Unexpected obstacles are unknown to the system. Vehicle reactions could be recorded by roadside mesh systems and information could be provided to later vehicles but this means roadside RF tags need to be more than just "tags." RF tag and vehicle system maintenance can be an issue. Finally people need to be able to over ride the system for various reasons most of which are safety. This however also means people will exploit the over ride for their own selfish interests.

 

In the past, quite a bit of research was performed on radar systems for road based vehicles. I don't think any of this worked out. One of the big problems was curves in the road. Poles supporting signs and guardrails look like road obstacles when the vehicle is going around a curve. If brakes were included in the control system, vehicles were always stopping quickly on curves.

 

Combining radar with the newer concepts? Perhaps, but I don't see many papers on this topic. I'm not sure why. Perhaps they don't provide enough benefit to offset the cost.