Synchronized S convoys for distance driving?

This is the classic "car train" theory for autonomous driving that's been floating around for a while.

I'm pretty sure it'll be coming sooner or later in some form. I would expect that dedicated train lanes will show up eventually in the same sorts of places that currently have HOV lanes - only a train lane can pack ~5x the cars into the same lane space, and still won't experience the stop and go issues.

Protocols and standards need to be developed for this - in particular all of the cars involved need to be able to talk to each other and exchange position, speed, and acceleration data at a minimum (ideally they'd share an understanding of the environment and all identified objects, so your car's situational awareness would extend the entire length of the train.)

Since the new Model S has all of the sensors it'll need and both 3G and WiFi radios, it should be possible to make it work - the rest is just (complicated, safety critical) software, and Tesla has a history of rolling out new features via updated software.

For ad hoc trains on the open highway, all you really need is the ability to pass the above data and programming for the ACC which causes it to follow much closer when it knows what the other car can see and is doing.

In principle if it knows the other car needs to slow down at the same moment the other car does, it could slow at the same rate and stop at the same distance it was following at - but being a foot or two off the bumper of another car at 75 mph might be a little unnerving for the passengers (most especially the so-called driver.)
Walter

Can't wait until this becomes a reality. I've been dreaming of autonomous high-speed car trains for long-distance traveling since I was a kid in the 80s/90s watching TV news specials about GM demoing the technology.

ChrisA said:

Can someone with more technical expertise than me comment on the following:

Last night unveiled autopilot, but not fully autonomous. Still, as Tesla gets closer to autonomous, it seems that the company at some point could add the capability for Model S's to essentially team up and drive a distance like LA to Las Vegas in close proximity so as to greatly reduce drag and increase range. Part of the question is really how close would a following car need to be to reap a substantial benefit and whether current capabilities of the sensors and software could soon be adapted to capture that benefit safely. The S drag coefficient is already low. At the same time, the wind resistance faced by a car is the cube of the wind speed, so a following car on a 75 mph highway out West may be able to very substantially increase range (even more so if there is a 10-20 mph headwind). A car traveling 75 mph using power typically required at 45 mph would a big deal.

Thoughts?

Click to expand...

The main problem I see with this idea is that the entire train of Tesla's would all arrive at a supercharger at the same time. It might work with two or three cars, but any more than that and the recharging time would start to get really slow. At least until they make the superchargers so powerful that two cars sharing a supercharger both get full power.

How much more range could this potentially give a Model S at highway speed??

I know that drafting behind a cyclist while cycling can save you 30% ++ in energy consumption.
If the same applies to cars, that would be 0.32 wh/mile * 0.7 = 0.22 wh/mile for the Model S85.

This would boost the range of the Model S85 from 85/0.32 = 265 miles to 85/0.22 = 386 miles.
And it would take the more Efficient Model S85D from 85/0.288 = 295 miles to 85/0.20 = 425 miles.

Imagine you drive on the highway in your Tesla, and you drive up to another Tesla. You ask the other Tesla to create an Ad-hoc road train. Both of the cars activate Auto-pilot and a communication connection. This allows the cars to have a really narrow gap between them. If the Auto-pilot or communication gets interrupted, a seperate safety system brings the cars back to a normal distance between them.

I really hope that Elon is thinking about this.

I have been a reader here for many years, but this is the thread that finally gets me to register an account!
Please give me your thoughts on this pressing issue.

wraithnot said:

The main problem I see with this idea is that the entire train of Tesla's would all arrive at a supercharger at the same time. It might work with two or three cars, but any more than that and the recharging time would start to get really slow. At least until they make the superchargers so powerful that two cars sharing a supercharger both get full power.

Click to expand...

Not an issue as long as there are enough bays available. Right now most Supercharger locations appear to have 8 lanes, meaning four sets of electronics and up to four cars getting full power at a time. So as long as the Supercharger was empty, a train of four could come in without creating any issues.

Tesla knows they'll have to keep adding bays as they keep more selling cars, and has money built into the plan to support that. If they add them in the same locations on the heavily trafficked routes, they'll be able to support larger trains.
Walter

The big problem is that it messes with other drivers who want to change lanes. If the cars could automatically read the turn signals of cars in adjacent lanes and increase the following distance when needed to allow traffic to flow freely, then I'd go for it. If not, we would need special lanes, probably in the far left, that serve this purpose.

Saghost said:

Protocols and standards need to be developed for this - in particular all of the cars involved need to be able to talk to each other and exchange position, speed, and acceleration data at a minimum (ideally they'd share an understanding of the environment and all identified objects, so your car's situational awareness would extend the entire length of the train.)

Click to expand...

I know that's what companies like Volvo are working on (seen some cool video demos), and the US Govt. is looking into standards of transmitting speed/direction similar to aircraft transponders, but...

...Why do we need those protocols and standards to do road trains?

Why isn't it as simple as follow the car in front as closely as possible at speed? As long as you've got the reaction times (and radar and sonar give pretty good indications of rate of change of speed very very quickly), that's got to be quicker than the lag over WiFi or bluetooth or any other secure radio based communications medium saying "I'm starting to slow down now" or "everybody panic stop!". And since you cannot guarantee actually receiving that message, or getting a transponder ping in time, you'd need the sensors anyway.

Please explain what I'm missing. Thanks!

Driving Miss Hazy: Will driverless cars decrease fossil fuel consumption?


This will probably answer most of the original questions of the thread.

woof said:

I know that's what companies like Volvo are working on (seen some cool video demos), and the US Govt. is looking into standards of transmitting speed/direction similar to aircraft transponders, but...

...Why do we need those protocols and standards to do road trains?

Why isn't it as simple as follow the car in front as closely as possible at speed? As long as you've got the reaction times (and radar and sonar give pretty good indications of rate of change of speed very very quickly), that's got to be quicker than the lag over WiFi or bluetooth or any other secure radio based communications medium saying "I'm starting to slow down now" or "everybody panic stop!". And since you cannot guarantee actually receiving that message, or getting a transponder ping in time, you'd need the sensors anyway.

Please explain what I'm missing. Thanks!

Click to expand...

The first answer is you don't - for a road train with longer following distances, which are still tighter than would be safe for any human. It's when you try to tighten up the reaction times and following distances that it matters. The protocols/standards would be to allow the car to recognize another car on autopilot and to get data from that car.

As fast as it is, reading off the radar that the other car is changing speed takes some time (you aren't transmitting a constant signal, so the worst case assumption has to be the a car starts to panic stop immediately after you bounce a pulse off of it; your programming may well require a few pulses to establish confidence that the car is slowing,) and the car will take some time to begin braking (tiny compared to the time it takes a person to recognize the situation and begin braking, but potentially significant with the ultra short ranges we're talking about.) How hard the other car is braking will take more time to figure out.

If the cars are talking to each other, the following car can get a "panic stop" signal from the lead car at the same time the car in front first sends signals to apply its brakes, shortening the reaction time by critical fractions of a second and thus allowing safe following at closer, more efficient distances.

The more extensive version with situational awareness would allow your car to see things that might affect it that the car in front saw but dismissed as not an issue for the first car - important since at very close following distances much or all of the field of view of the front radar, sonar, and camera is blocked by the lead car.
Walter

Doesn't a road train mean the car in front will be the same as any other car? Who pays for that difference? Why would I want to be at the front of a road train..?

tom66 said:

Doesn't a road train mean the car in front will be the same as any other car? Who pays for that difference? Why would I want to be at the front of a road train..?

Click to expand...

I don't think so. The car in front should have some drag reduction because the low pressure area behind it isn't low pressure any more. It probably won't get as much benefit as the cars behind it, though. I'm not sure how one would decide who should be in front - and as long as there are enough superchargers everywhere you're going, I'm not sure that it matters much whether you train up on normal freeway lanes.
Walter

There is a drag reduction for the first car, but geese figured this out some time ago for their long distance tailgating travel. The leader just falls back when (s)he gets tired and the next in line moves up to the point.