This is already done, effectively. Freight trains often move pretty slowly because it's efficient, even though obviously the technology exists to make trains go really fast. According to this (if I understand what it's talking about correctly), the average speed of a freight train in the US is barely over 20MPH:

http://www.rita.dot.gov/bts/sites/rita.dot.gov.bts/files/pub...

I don't know what, if anything, this implies for autonomous truck speeds, but it's something.

The average mainline long-haul speed of a freight train is probably around 50mph. The average speeds of freight trains are remarkably lower since trains spend a lot more time stopping (if you're dealing with local freight, you're going to be stopping every few miles to detach or attach a few train cars). Looking at <http://www.railroadpm.org/Performance%20Reports/BNSF.aspx>, the intermodal trains (which tend to be between major shipment terminals and not including local businesses) are averaging at ~30mph.

Is that because the slow speed is efficient, or because it's cheaper to avoid improving the track to support higher speeds?

Freight trains in Europe are much faster[1] -- but that could be because they often need to fit in with the schedule of passenger trains using the same lines (which themselves require the track to be maintained to a high standard).

[1] Citation needed, I only have observations of them going at a decent pace through stations I'm waiting at.

The train networks in the US are more optimized for, and better at freight traffic, while Europe is more optimized for, and better at, passenger traffic

http://www.coyoteblog.com/coyote_blog/2008/02/european-vs-am...

That's probably exactly why freight trains are faster in Europe:

http://www.economist.com/node/16636101

One really nice anecdote from that response:

> In fact, the freight rail system is so efficient that to some extent we've obviated the need for the Panama Canal. Many Asian container ships bound for Europe actually make port in Seattle or Vancouver, offload their containers onto trains which shoot across the country to New York or another eastern port where they are reloaded on ships for the trip to Europe.

That is impressive.

Though I suspect soon enough the Northwest Passage will be clear for enough of the year that most freighters will just sail through the Arctic. Perhaps both Panama's government and North America's railroads should be pushing hard for a carbon tax.

Good question, I'm not sure how much speed plays a role in train efficiency. It's no doubt some of each of your alternatives, I imagine. Even the higher speed ones are more like 60-70MPH. Nobody seems to be the least bit interested in building a TGV freight version, for example, even to run on existing TGV lines.

[1] has rolling resistances for railways and cars on roads, which are almost an order of magnitude apart.

Intuitively, I'd expect that once the train has accelerated, it will keep its momentum very nicely, and (if electric) can easily return energy if it does need to slow down.

I can't find a real-world coasting distance for a train on level track, only reports during accidents/near misses -- 5km under emergency braking on a slight downhill slope, for one! [2]

[1] https://en.wikipedia.org/wiki/Rolling_resistance [2] https://www.gov.uk/government/uploads/system/uploads/attachm...

Rolling resistance is roughly constant with speed, so it's unimportant for efficiency. The main driver of changes of efficiency with speed is aerodynamic drag, because that increases with the square of the speed. Trains presumably have very low aerodynamic drag because they're long and thin.

The calculation is straight forward. The coefficient given for rolling resistance is 0.0004 for rail which means that .0004W is the frictional resistance of the track. Given a train going at velocity v then the train is accelerated by friction at a rate of -.0004W/m (F=ma so a=F/m) which is equal to -.0004g (W=mg).

Then to stop it will take t = v/.0004g which means that you will travel vt/2 in that time.

For example, if a train is cruising along at 30 mph (44 fps) then it takes 44 ft/s / .0004(32.2 ft/s^2) = 3416 seconds to stop which means that the train travels 14.2 miles under ideal conditions.

I suspect that the average speed of a moving train is higher than that, though. (In the same way that you can't take distance to the office divided by time to get there and conclude anything about average driving speed, due to time at stop lights, time in stop-and-go traffic, etc. When you're actually moving, you're moving faster than that statistic indicates.)

I also wonder if we're going to see an increase in traffic due to zero-occupant traffic. Supposing that the car of the future can drive itself home when I'm at work and pick me up at the end of day, will that mean that there's going to be a lot of cars on the road just transporting themselves and nothing else? This might be particularly common in cases where a single car is shared by multiple people.

I think individual car ownership will disappear. You will commission a car to get from point A to point B (with the options of lower costs if you are willing to pick up passengers en route) and then the car will go on and do other trips. Your own car just sits there almost all the time, that's really inefficient and expensive.

Eliminating car ownership makes a lot of sense in any urban area, but I bet people will continue to hang on in rural and some suburban areas.

In rural areas, an autonomous vehicle would need to travel long distances to pick up and drop off folks. It could be prohibitively expensive or inconvenient. People also use vehicles for other things in these areas, both work and play, which do not fit into the taxi model.

In suburbia, the taxi idea could work, but there might be advantages to owning your own vehicle. Maybe a family will own one autonomous vehicle, just one, but it would be in constant use throughout the day: mom goes to work, dad picks up donuts for coworkers, son goes to school, daughter goes to soccer practice, son goes to track practice, dad goes to get groceries, mom goes to get dinner, son goes to friend's house, daughter goes to football game, mom goes out to gym, etc. You'd have your one family car zipping back and forth between all these events in the family, picking people up, dropping them off, going off to do something else, and coming back again. Having an autonomous car in the family could make sense depending upon the costs.

And this is why Uber has such a crazy valuation.

How did you reach that conclusion?

If driverless cars are a "thing" in the future then what is Uber's USP? Why would you use Uber rather than one of a dozen other companies? Is it just their capital advantage? Their network of now unemployed drivers? Just the name and nothing else?

What is it exactly that keeps Uber in business with driverless cars?

The switching cost between apps is practically zero so if someone else manages to figure out a better routing / experience with picking up people (some people use UberPool for dating, really creepy shit!) etc then Uber will become a footnote in history quick. It's now hard to disrupt Uber in its own market but when it's a new market?

Uber raised their funding on the basis of their current business, not autonomous vehicles not any other delivery/logistics plays.

In bigger cities, if you compare this to cars looking for parking spaces, I think you'll actually see a net decrease in traffic.

* Why would it have to go home? It could just drive out of town to any cheap loitering lot.

* Why not put it to work. It could transport people or packages during that time.

* Mostly, in such a future, you probably won't own a car. You'll rent or lease access, and the price to take a nice autonomous mini-bus will be super low while remaining convenient.

Seems unlikely. The cost of electricity is pretty low, while the cost of utilizing your truck is probably much higher.

That'd be interesting and it seems to make sense. Since business's optimize for cost and AI driven trucks don't need breaks and won't have limits on driving for extended periods of time (presumably) it seems it'd make sense to optimize for electricity use and delivery times. I don't see why they wouldn't drive as slow as they needed if that would produce greater profits for the shippers while not reducing the end consumer experience with late deliveries.

Oil tankers optimize for both fuel consumption and timely (= at the right moment for the spot market) arrival. If prices are low, tankers can get 'parked' for months. If that's the case at one's destination (or rather: possible destinations), there is no need to rush there.

I also think truck drivers already adjust their speed depending on economic reality. If, say, you are transporting fresh flowers towards an airport, arriving in time is more important than when you are transporting furniture to a furniture shop.

That certainly is true at the extreme ends. Drivers who drive to be transplanted organs between hospitals get special training to drive fast, while drivers driving garbage to a landfill likely will not be pressed for a timely delivery (although I do not rule out that someone who knows the latter business will tell me that missing one's reserved spot for delivery at a landfill can cost a lot of money)

Or because the warehouse they're driving to hasn't made space for its cargo yet?

A lot of cargo ships have been doing half-speed for a couple years to save money and produce less pollution.

Minimum speed limit? As far as I'm aware, in the US you are supposed to go the posted speed. You could get a ticket for going slower if you are impeding traffic trying to safely go the posted speed. Are there roads in the US that specifically allow a range of speeds (beyond the variation seen because human drivers don't follow the law)?

> Minimum speed limit? As far as I'm aware, in the US you are supposed to go the posted speed.

The posted speed is usually an upper bound (though in some cases only a presumptive one, not a conclusive one, the way the law is actually written.)

> You could get a ticket for going slower if you are impeding traffic trying to safely go the posted speed.

You can get a ticket for impeding traffic, but usually going slower than the limit is not sufficient for that. (And may not even be necessary.)

> Are there roads in the US that specifically allow a range of speeds (beyond the variation seen because human drivers don't follow the law)?

Every highway (which is much broader than freeways) in California that doesn't have an expressly posted minimum speed limit, plus every one that does have an expressly posted minimum speed limit that isn't identical to the posted maximum limit. (California Vehicle Code Sec. 22400 allows a minimum speed limit to be established and posted -- which is still, where established, a soft limit -- and provides a violation for impeding traffic with or without such a limit, but does not set the posted maximum speed limit as a minimum speed limit, soft or otherwise, where no minimum speed limit is posted.)

Part (a) of VC Section 22400 is what I was remembering from way back when I took driver training, and it states "No person shall drive upon a highway at such a slow speed as to impede or block the normal and reasonable movement of traffic, unless the reduced speed is necessary for safe operation, because of a grade, or in compliance with law."

Wouldn't driving below the posted speed limit when there are no hazardous conditions fall under "as to impede or block the normal and reasonable movement of traffic"?

> Wouldn't driving below the posted speed limit when there are no hazardous conditions fall under "as to impede or block the normal and reasonable movement of traffic"?

Not normally, driving below the posted maximum speed limit would not, in and of itself.

It might -- and, given the soft nature of upper limits, it might even without driving below the posted maximum -- in the presence of other traffic, depending on the road (what will impede traffic on a single-lane road is different than what will impede traffic on a multilane road) and particular traffic conditions.

I understand that, but I guess I'm of the opinion that while we provide leeway to humans because of our nature, an automated system that drives below the speed limit in the case that it is blocking or restricting traffic be reducing the number of lanes flowing at full speed really has no excuse, other than purposefully slowing the normal flow to save money. I imagine that might be successfully litigated.

Could/would they drive well below the speed limit without legal problem when no traffic is around? I imagine they could, but I'm less interested in that (but understand my question was ambiguous).

Sorry, let me confirm this. You can get a ticket for driving slower than traffic, despite driving at the speed limit?

Could you, in the same situation, also get a ticket for following the traffic—i.e. breaking the speed limit?

> You can get a ticket for driving slower than traffic, despite driving at the speed limit?

Some speed limits in California are soft (it depends which of the provisions allowing speed limits they are adopted under, or whether they are the default limits.)

It is legally possible for traffic to exceed the speed limit where those limits are soft without necessarily being illegal, unreasonable, or safe.

It may, in some circumstances, therefore, possible to impede reasonable flow of traffic while driving at the speed limit in those cases, thus breaking the law.

> Could you, in the same situation, also get a ticket for following the traffic—i.e. breaking the speed limit?

Yes, but if the circumstances were such that someone could be impeding traffic at the speed limit, it would also necessarily be true that it would possible for you to exceed the speed limit and not be breaking the law. But the burden of proof would be on you to show that the driving was safe.

Yes, and in states with a 'basic speed law', or similar, you can even earn a ticket for driving substantially below the posted speed limit. The Utah driver's handbook and license test, for instance, explicitly call out that road conditions may be such that the maximum safe speed is 0 mph.

Massachusetts Turnpike has a minimum speed of 40 unless you are in a tunnel where the minimum is 20.

http://www.mit.edu/%7Ejfc/NMA/730CMR.html

Most of I-93 in NH has a posted minimum of 45.

Some states (forget which I was driving through) have posted minimum speeds on highway.

Others I'm sure there is a generic lower limit "clause".

I've never seen a state that didn't have a "at a speed safe with current rode conditions". So, that you can still get a ticket going 65 in a 65 but there happens to be a blizzard going on.

I doubt any state requires you to go exactly the posted speed limit.

There are cases of people getting obstruction of traffic tickets for going at the speed limit. It happened in Toronto. One guy got a speeding ticket in the merge lane, so in a pique he did the posted limit in the fast lane, which normally moves 15-20 mph faster than the limit. He caused a traffic jam and got an obstruction ticket. He fought the ticket and lost. Sorry, can't find a link now.

Many roads also have signs and states have laws along the lines of stay right except to pass. So if you're not actively passing in the left lane, you're breaking the law in many places even if you're going at the speed limit. I expect few tickets are given out for this even though it causes dangerous behavior.

In Missouri, we have signs posted where the Speed Limit is 60 or 70 and the Minimum is 40.