Hacker News Reader

Comment by tyrw | Hacker News Reader

All of the comments re: swappable battery packs are missing the end goal. Once the driver is gone from autonomy, these fleets are going to be dirt cheap to operate.

It won't make sense to swap a battery pack when you can just park one truck on the charger and let another take its place.

View on HN

I'd also add that virtually the only compelling reason to have a single, large load like a big rig is to minimize the cost of a human driver relative to payload shipped. Without a driver, the optimal load size will likely be much different and will vary more by industry than it does today.

If anything, wouldn't the loads get larger.. since you could build trailers with more sensors, allowing more trailers to be attached to the same tractor... minimizing energy/fuel cost.

One tractor carrying larger loads is going to be more efficient than many small trucks with smaller loads: total weight over payload.

A human being can only safely handle so much cargo, and that's probably lower than could be automated.

The limit on truck size isn't because of human driving capacity, it's because road damage scales as the fourth power of axle weight. Trucks already do the vast majority of damage to our infrastructure, increasing the weight more would literally take down some older bridges.

I was referring to more trailers like this: http://www.hankstruckpictures.com/pix/trucks/macneil/2004/ma...

That doesn't require an increase in axle weight

I seem to recall hearing that some states have restrictions on trailer count. Presumably due to road safety issues.

Hmmm, I'm not sure the equation is so obvious.

If you don't have a driver, then the vehicle amounts to just a couple motors, batteries, and sensor arrays.

It doesn't make a difference if the battery is located in the front "truck" part of the convoy, or whether it's distributed throughout the trailers.

For a very long chain of trailers, you probably want a set of motors for every trailer right?

And as you've speculated, there's going to be a sensor array in every trailer.

So basically, you have motors in every trailer, batteries in every trailer, and sensors in every trailer. At that point ... why do you need a "truck" on the front and/or why are they connected at all?

Dropping a trailer vs waiting for it to be unloaded? I dunno. I agree with what your saying. Just moving stuff is obviously faster than waiting around for loading and unloading. At what point does the idle electric truck start to cost less than a simple trailer.

There's an "easy" solution to that too, you make it a flat bed electric self-driving trailer which takes a container similar to sea shipping containers on top.

Then you don't:

- have the unnecessary truck cab in front

- cost of the trailer being unused while loading / unloading

- potentially drastically change shipping containers received from over seas.

good point.. but that depends on the cost of the sensors, motors, etc. You could have a reduced set of sensors on the trailers, and more advanced set on the tractor.. and that may be cheaper than adding advanced sensors to every trailer.

Modular self propelled trailers are already a thing for moving heavy equipment they're just not for road use so most of HN will never see them.

You usually have a 100-500hp diesel running a big pump and hydraulic motors at the wheels.

Yes and no. Cost of fuel is also a big factor. A big rig is typically running about 6 MPG, and for lighter loads that doesn't change very much if you're hauling a half-size or full-size container. Sending two trucks means burning twice as much fuel, which could easily be hundreds of dollars.

But definitely that's less of a factor with an electric truck since the charging cost will be a lot lower than a comparable amount of fuel.

Cost of fuel will also be greatly reduced for the use cases within electric range:

- Removing the driver allows for more aerodynamic cab configuration, or no cab at all

- Electric motors are more efficient, particularly in high-torque / low-rpm environment

- Massive & inefficient 18-speed gearbox can be avoided

- Cost of fuel itself (electricity) is much cheaper

I think the bigger savings will come in being able to precisely control stops, starts, and turning, which will both minimize damage to the road surface and lengthen tire life.

ISTR a case where a large open pit mine bought GPS-driven trucks and saved a great deal just from longer tire life (each tire was $20K).

Re: gearbox, I wonder whether an autonomous truck could be paired with a little scouting drone to "look ahead" at things like blind corners or stop-and-go traffic, to allow the truck enough foreknowledge to gear up and down more gradually.

It is almost certainly cheaper to have big motors centralized at a single unit, because motors scale faster than linearly with size.

It is also almost certainly cheaper to have small batteries spread through the train, because batteries scale mostly linearly, but there are a few problems with dense packs (like cooling).

Thus, if you want to plug the unities one into the other, you will probably have one or two powered unities, many cargo unities, and a few battery unities spread between them. Besides, you will also want dummy unities on the front and back with aerodynamic shapes.

Capital costs? If a truck can only operate 50% of the time because the other 50% is spent recharging, you'll need twice as many trucks to move the same load.

While those trucks are recharging you're still paying for them.

Depends on how expensive the battery is compared to the rest of the truck.

>It won't make sense to swap a battery pack when you can just park one truck on the charger and let another take its place.

That depends on the relative capital costs of a truck sitting idle for hours while it charges and the costs of the systems for swapping and charging batteries.

It's the difference between making every truck and charging station a little more complex and making every station significantly larger and increasing the number of trucks you need to own.

It's still a hell of a lot cheaper to have 2 batteries and 1 truck than 2 batteries in two trucks, assuming switching cost isn't too high.

> It won't make sense to swap a battery pack when you can just park one truck on the charger and let another take its place.

I'm guessing an operator would want 100% utilization from a truck.

If a battery discharges at the same rate as charging, then you would need 2x the number of batteries. You would need more batteries if charging is slower than discharging.

More accurately, an operator will want lowest cost of operation. You can add complexity, cost, and failure modes to every single truck by making the battery swappable, then build infrastructure to swap the batteries, or you can just build more trucks. Given that wheels and axles are likely to be the cheapest part of the equation, I'm betting they'll buy more trucks.

I remember hearing there is not enough lithium in the world for all of Tesla's ambitions. They're hoping for battery revolutions

I found an excellent Quora answer that dealt directly with this. Well worth a look-over.

https://www.quora.com/Is-there-enough-lithium-in-the-world-t...

tl;dr- Lithium supply (in terms of raw quantity of element available) is not going to be an issue, ever. Economical extraction might be.

And you can make batteries out of other materials. At current prices lithium might be best, but more demand will lead to different prices; and if lithium supply doesn't pick up, people will substitute.

> Once the driver is gone from autonomy, these fleets are going to be dirt cheap to operate.

What gives you the impression that human drivers are such a large part of the costs now?

I don't know whether it counts as "such a large part of the costs", but at least in the US it does appear that wages and fuel are about equal as the two major costs:

While fuel costs were historically the largest share of total cost, driver wages surpassed fuel as the largest share at 31 percent of total cost in 2015, and fuel now accounts for 25 percent – its lowest share since the inception of this study. Fuel was followed by equipment lease or purchase payments at 14 percent, with repair and maintenance, insurance premiums, permits and license, tires, toll costs, and driver benefits each representing 10 percent or less of average total marginal costs.

http://atri-online.org/wp-content/uploads/2016/10/ATRI-Opera...

Tables 10, 11, and 12 show a more exact breakdown. If all else remained the same, removing the driver would appear to reduce total costs by 40% (30% wages plus 10% benefits). So unless the electricity is much cheaper than diesel it won't be "dirt cheap", but certainly would be a significant reduction.

Reducing costs has a large effect on profits. If your profit margin is 5% then reducing costs by 10% triples your margin.

Removing the driver will also help with capital costs: trucks running 24/7 can spread those out over more hours.

Not if those trucks must stay most of their time parked for charging.

Reply on news.ycombinator.com