Autonomous trucking is harder than autonomous rideshare
kevinchen.co
kevinchen.co
(I used to work for an autonomous truck company, and when you factor in the cost of roads in addition to the development of the trucks, it makes absolutely no sense to do autonomous trucking when you could do trains. As a culture, we've been brainwashed not to fund trains. We collectively spend billions and billions on roads but would not dare spend money to build more tracks. It is shocking and ludicrous, but that's what happens when you suck up a century of propaganda from the fossil fuel and automotive industry).
Here's another example - look at all the little rail branch lines servicing parts of 1944 Boston: https://ngmdb.usgs.gov/img4/ht_icons/overlay/MA/MA_Boston%20...
I don't know how you bring that back after decades of building for cars.
What I did not appreciate, but they did, is that these rails can bring an absolutely astonishing amount of noise into a town, and behind that a long trail of associated social ills. There is a gradual sifting of residents within a certain distance of these trains, based on who is either loud themselves and thus doesn't mind, or else must tolerate the amount of noise these things make.
Our town is old enough that some of the 'nicer' neighborhoods predate the introduction of the rails, so there isn't a strong confounding signal of 'bad' neighborhood correlation at work here. In at least one case there was a house we pushed past our agent's scheduling machinations to see for ourselves, and would likely have closed on had we not happened to visit at just the right time, when a multi-engined repositioning train (which we had no idea was a very common guest on what we thought was a mild mannered commuter rail line) came through. Though unseen and multiple blocks away, it still shook the floorboards as it passed.
Both older DC and newer AC trains are effectively silent when coasting on a straight
The main sources of noise are cornering and braking. I live in a rare city that has narrow gauge for all of our rail, and am less than 200m from the busiest corridor in our state
I regret chosing an apartment on a tight bend (narrow gauge's bends are especially loud)
But if I were to live 500m away parallel to the track (as in, still 200m away from the track) the mostly AC electric trains going by every few minutes would be largely inaudible (<40dB)
My main point is tracks aren't loud, tight corners and areas around stops can be loud
I fear city planners aren't testing their decisions closely enough, though
Anyway, noise pollution has a real impact on health, but so does particulate carbon and benzene from freeways. Before moving near a major road, find out what sort of PM 2.5 is typical, and check to see if everything is coated in black dust. If so, and you can afford some other place, move there instead.
A well maintained line and fleet would also not have that much screeching.
I love the noise and the trains. You get used to the sounds, and it becomes comforting. Like a loud rain on a rooftop. I feel strange when it's quiet.
There are some homes by Georgia Tech that are literally feet from the tracks [2], and I think those are awesome. I also hope these trains don't derail frequently like the ones in Hulsey Yard do.
[1] https://m.youtube.com/watch?v=ZVX0KMTa8ZA
[2] https://www.redfin.com/GA/Atlanta/576-Marietta-St-NW-30313/h...
Just require distribution centers or warehouses to require a rail connection to the state's rail infrastructure to be approved for construction.
If no rail connection exists, then the state will negotiate who pays for what to extend the network to it.
1. We actually do not have a market mechanic that exists in the United States that allows people to choose properly dense homes or neighborhoods because of a number of reasons. Search tools and things like that focus on metrics like square footage or number of bathrooms, and so people select for that "bigger is better" "more is better".
2. State departments of highways (they don't do anything besides that in most states) consist of individuals and contractor companies who have become adept and normal with building highways, and so all of the organizational inertia is on building anything that looks like a highway for the safety of drivers, to the detriment of everyone else. Since only highways get funded and built in most of the country, we end up having other companies like fast-food restaurants build in ways that cater to the driving demographic.
I imagine that rail line makes a big difference in being able to keep up with the logistics advantages of 6 big box stores within 30 minutes.
Seems US is doing pretty well on prioritising freight rail transport, despite brainwashing, at least compared to europe.
Which is probably a good thing. Until that figure is a lot higher, passenger rail is just one more thing reducing the efficiencies of freight.
In Canada, to my understanding, it's the other way round where passenger trains have been reduced because of the need for more cargo train trips. The USA might be similar.
But the USA is VAST. Freight trains are LONG. Not much of America's West is actually double track. And when you have a shorter Amtrack and a long freight train the Amtrack loses. The Amtrack has to wait until the freight train has passed.
As the freight trains run on PSR [1] there is no fixed time as to when they pass and thus the Amtrack can not run reliable.
As such it is easier for me to drive from West Los Angeles to Las Vegas, or even to drive to LAX and then fly to LAS than it is to take the train.
https://www.bts.gov/newsroom/2017-north-american-freight-num...
https://www.aar.org/facts-figures#:~:text=Freight%20rail%20a....
I’m unclear why you’d care about value at all in the context of this discussion. In context weight and volume are the most relevant metrics.
Travelers and commuters win, cargo wins!
It is funny though, the US built it's current freight network in a comparatively low tech era, yet I doubt it could do it again in the current era due to funding and beauracracy.
Also, have you heard of the Alameda Corridor is Los Angeles County? That project is incredible. The amount of soil moved for that project must be staggering. The dimensions: <<The centerpiece of the new Alameda Corridor would be the "Mid-Corridor Trench" a below-ground, triple-tracked rail line that is 10 miles (16 km) long, 33 feet (10 m) deep, and 50 feet (15 m) wide.>>
No one (generally) bats an eye when voting at the city level for a $100mm road package.
But trains?
Freight should move mostly by train. We could build rail lines where there are already highways, just swap a vehicle lain for a train rail, all that 18 wheeler traffic can be reduced significantly.
Safer highways. Less congestion.
Rail truly is a unique infrastructure concern and needs to be designed and laid on its own.
Passenger trains typically (want to) run faster than freight, so the requirements are stricter.
Caltrain tracks and right of way was purchased in 1991 from Southern Pacific (now Union Pacific) by Peninsula Corridor Joint Powers Board which is basically Muni, SamTrans and VTA for most of the route.
Here is a detailed map of who owns what https://caltrain-hsr.blogspot.com/2009/01/caltrain-right-of-...
While it is true that Southern Pacific owned the line for most its history it was always built as commuter line. Union Pacific is only allowed to operate 3 freight lines per day on weekdays now.
Freight does not take priority, historically freight did not fund the development of the line, Characterizing it a freight line with passengers added on to it is not a fair representation.
--
I cannot speak for Amtrack, I don't know the network personally and it is also big and very composite a network to make any general statements like for Caltrain. However in general commuter heavy lines were built as commuter lines very early one because there was economic case to do so. It is the only the long distance lines which do not have the economic viability on their own that can only run on freight lines.
...or just kill everyone, problem solved :)
I don't know what building a second parallel railroad system to take over this task would cost, but I'd guess several Iraq Wars.
Making the trucks autonomous doubles their utilization, which means we need half as many trucks, and makes freight cheaper, which makes prices lower, benefitting everyone, especially the poorest.
> As a culture, we've been brainwashed...
This is never the start of a insightful argument!
Compare also https://en.wikipedia.org/wiki/Impact_of_the_privatisation_of... and see eg the share of rail in all transport.
A rail line is single use, single destination. commuter and cargo cannot mix (one goes high speed and one goes super slow) if they do mix it shuts the entire segment of the track down until it’s off that track.
Roads are multi use , multi location. It seems unfair to compare the cost based on road maintenance vs track maintenance. I would expect something to account for the sheer difference in volume each does
Road trains are both intermediate solution everywhere and also can operate in routes where trains are not economical enough, even if there was a will to do it.
The billionaires are surely investing heavily on trains and farms.
Which goes to show that money doesn't have that much of an influence in politics, despite popular opinion to the contrary.
The cab can be a lot cheaper, if you don't have to keep a human comfortable and safe inside. Also keep in mind that an autonomous truck can drive 24 hours a day, and doesn't need to take regular breaks throughout the day nor sleep. (They will need maintenance, but probably not more than a manned truck.)
You are right that the costs are real. Things like first/last mile (or loading and unloading) would probably need a major reshake of the industry, if the truck doesn't bring its own labour, in the form of a driver, with it.
Amortized over the life of the truck they really aren't.
Heck, maybe they don’t even need to be articulated. They could look like slightly larger Amazon or UPS delivery vans. No need to worry about jackknifing.
If it’s a fleet maybe battery packs can be swappable so when it arrives at a hub to load/unload it can pick up a charged battery pack.
The only issue is for cargo that can’t be split up.
> Therefore, all safety-critical decisions must be made by the onboard computer alone.
Why is the requirement that all safety-critical decisions must be made on board, versus the seemingly-simplifying assumption that only some or most decisions would have to be made on board, because a remote operator or backend service could be available a lot of the time? It doesn't seem unreasonable to me to have a single operator remotely monitoring multiple vehicles that are autonomous under ideal conditions (driving along a straight road in good weather) and then taking over when necessary. Let's say you would only use such a system for major routes with solid satellite visibility, not last-mile routes hauling heavy equipment on a dirt road in the boonies, or something like that. Maybe this wouldn't work, but it's not obviously ridiculous to me, so I wonder why he just starts out saying the truck most be fully autonomous with no human ever in the loop.
Yes, and that has to be safely possible. From my reading, the article has no problems with remote operators giving command. The truck just also needs the ability to both safely execute them and come to some safe state by itself if necessary commands don't arrive [in time].
The interesting question is how complicated those safe states are to realize. Just hitting the breaks on a freeway seems hardly acceptable. Generally, this being harder than for cars in cities is a core conclusion of the article.
As an additional safeguard, you can make your trucks go into 'safety' mode when connection becomes spotty or when too many operators become too busy with other trucks.
'Safety mode' could mean slowing down the trucks or even stopping some of them. And in general, letting the autonomous systems err on the side of caution more often.
Regarding the minimal risk condition / fallback behavior, a central point of the article was that slowing or stopping are almost always unacceptable on freeways because of the speeds involved
Unlike autonomous cars where everyone needs to go wherever, inter city trucks have fixed well known routes with predicable volume of cargo that can be easily chained together.
Bulk freight trains solve that problem already. Simply use flatcars or intermodal containers then use rubber tyred trucks in the last kilometre.
Rather, things like "unknown obstacle on the road, can't continue", with the operator instructing it to just run over that empty cardboard box. All realtime safety decision-making should have safe autonomous fallbacks, a human cannot be expected to drop in and take charge in seconds. The entire level 3 / 4 "semiautomation" market is a bunch of make believe.
If the AV ran into an issue, it could tele-help and have a human operator take over.
Completely makes sense that you could have a person monitoring multiple AV trucks and take over driving when needed.
This analysis seems really suspect to me. Any clarification would be appreciated.
Also recommend checking out the citation. It is an accepted value used in American highway design.
So between which endpoints shall you measure "reaction time"? I'm already priming my reaction by actions which lead up to it. Obviously, an automated driving system doesn't need some of these steps, but it's still running through the same scenario and "thought processes" to reach a decision and course of action.
Once at lunchtime, I had two coworkers in my car. I was driving a manual-transmission Acura Integra in light rain, heavy traffic in downtown Palo Alto. I had a left turn signal and I proceeded through the intersection, when an oncoming car was coming straight at us and couldn't stop. I punched the accelerator and sped through the turn without incident. My coworkers both congratulated me on making the correct decision: to hesitate, or to hit the brakes, would've surely crashed us. But what was my reaction time? My foot already on the gas, I simply made a snap decision in the moment to follow through.
There are a lot of differences between passenger vehicles and trucks. The physical dynamics of articulated vehicles, the mission profile, and social dynamics come to mind. How does a robotruck place cones or flares while it awaits rescue?
Personally, I expect autonomous trucking to be a force-multiplier for humans who were formerly drivers. Such trucks will have sleeper cabs and the human will be there to maintain the vehicle and handle the long tail of tasks (filling tires, cleaning, refueling, repairs, rigging, whatever). You'll get 24-hour operation out of a single human employee because they'll be able to sleep and do other things most of the time. Maybe they'll work a second job as a remote call-center operator.
For lidar, the range is also limited by power limits + physics, which cannot be overcome by increasing money/power/device size. Some dependencies on semiconductor manufacturing tech or better signal processing might be possible to solve with more money.
Short-haul trucking (in-city delivery etc) is where the value is in having agile, self-driving vehicles moving cargo away from supply points… but it’s also the hardest to implement because there are so many dynamic elements in city driving. And largely these light delivery vehicles are already being electrified where it makes sense (UPS, USPS, etc). You are mostly talking about it any advantage coming from removing a human, not the act of rolling electric vehicles into it like taxis where it’s primarily disrupting internal combustion with electric.
Also, any system is eventually going to be utilized at its full capacity. The number of vehicles supervised per agent will increase until the average number of incidents at any one time meets the average number of agents, or beyond, until there is a regulation. Capitalism is a cost optimizer and no cost will ever go unsqueezed - much like the internet is designed to route around network damage, capitalism is designed to route around any such inefficiency as morals or safety standards until otherwise compelled by regulation.
You say it like this is a point against autonomous trucking?
Technical problems are solvable. Political and social problems are basically intractable in practice.
If autonomous driving can turn 'delivering large amounts of cargo along fixed, guided pathways' from a political into a technical problem, then this would be a jackpot.
> [...] capitalism is designed to route around any such inefficiency as morals or safety standards until otherwise compelled by regulation.
No? Plenty of eg cars exceed minimum safety standards. Some brands, like Volvo, are even explicitly sold on safety. Most companies explicitly talk up their morals and ethics, too. Look at almost any old advertisement for examples. Ethical brands are quite popular, and whenever one is found to be only pretending, there's usually a big scandal. So many of them actually practice what they preach.
Similarly, most people are paid more than regulated minimums. According to your theory, that shouldn't happen.
In the context you're discussing, that form of safety isn't an inefficiency but a marketable feature. That doesn't mean that safety in general is a marketable feature.
Those safety standards tend to omit the safety of any pedestrians being hit - otherwise, cars would be far lighter (and with bonnets far lower), because a heavier car benefits the occupants in a crash at the expense of whoever they're hitting.
This is a video from our firefighters here in slovenia... Humans are shitty drivers, sure, but how the hell would a computer react to a chaos like this?
Also, since it's a relatively time-insensitive situation (as in, tens of seconds to deal with it rather than seconds), once it identifies that it's an exceptional situation, it can have someone connect to the truck remotely.
And get overtaken by other drivers :) Then a lot of honking and swerving would be needed to get back in the correct lane. Also they sometimes need to back up a bit, to let a 'late' driver to move infront of them, or behind them, move to the other side, interpret what honking means to whom, etc... or drive forwards a bit, or whatever the firefighter knocking on the window tells them, speech recognition is still not perfect and interpreting abstract orders is even worse.
This is covered in TFA, and noted that one cannot rely on wireless connectivity along all highways. Certainly not immediate responses, and doubly-so when the vehicle may be in a deadzone; either a natural one or one caused by overloading the local towers (such as one might find at a remote accident with lots of other motorists) will do.
Focusing on specific routes applies to many many businesses too.
The ability to provide more human assistance in tricky situations doesn't apply to trains at all.
Your claim that they are "describing trains" is utter nonsense.
Also, I feel like there's a lot of talking past one another in these conversations because one person will say "Let's see an autonomous truck shipping hazmat to Pittsburgh in February with freeway lanes shut down" and another person might say "that's a rare instance" but I really don't feel like society will accept anything other than trucks / vehicles that are able to operate under all conditions, with greater safety than the safest human driver. We tolerate human failures but to use them as the benchmark for autonomous systems would be perceived as unethical, because autonomous systems are deliberately designed and any failures by them would be seen as an intentional oversights and errors, and no one at Waymo or Tesla or where ever is ever going to be charged with vehicular manslaughter for an autonomous vehicle error. We'd demand a way higher standard because these companies don't really have any skin in the game, except for financial penalties which we now understand is not a deterrent for anything. My observations are only moderately related but I'm anticipating the same well-trod talking points coming up and want to address them.
That’s not the argument being presented though. For example Waymo claims to exceed human performance by a large margin: https://waymo.com/blog/2023/12/waymo-significantly-outperfor...
(Again, one may disagree about the methodology or the conclusions of the study. Just want to point out it’s not the argument being presented.)
For now what you could do is to demand that anybody that has a DUI or other such item on their record to mandatory only be allowed to be in vehicles that have self driving if the manufacturer is willing to assume liability. And if that doesn't happen then they might as well take a regular cab.
What happens when it snows ?
I am all for autonomous vehicles for other people. And you can call that elitist and I frankly couldn't even argue with you, but god damn, so many people have utterly no business being in control of a car. I travel quite a bit both for work and leisure, almost always by motorway, and some of the shit I see is just mind-bending. So much inattentive driving, so much pointless risk-taking, as they say it's time for Wisconsin's favorite game: Why's that car being driven badly? Old, stupid, drunk, or all three?
And like, there's not really an answer for a lot of this. As much as I love them, cars suck on a society scale. They're basically a poor tax as even the most low-income people in my neck of the woods must have them to get around, there's a bus network but it's shit and it has very low ridership, and that's all my city has apart from taxis. And, old people use them for the same reason, even if they're damn well aware that they shouldn't really be driving anymore, if they don't have someone to get their groceries n such, what are they supposed to do?
So yeah, self-driving cars are awesome. In theory. But then I see video of someone using the lane-guidance on a Tesla where the car just sees... who knows, something, and suddenly jerks to the right, right at pedestrians. Or we get the stories of autonomous cars just shutting down and refusing to move, even for emergency vehicles. Still others, they end up parked on someone's legs, and the stupid support system can't allow a remote operator to move the vehicle off said leg.
At this point, as much as I hope we get it, I also kind of recognize that truly autonomous cars simply might not be viable yet. I think a much better idea is to just de-carify our societies. I don't want them gone entirely, but people need more options than JUST buying or leasing these enormous, polluting, dangerous machines that they do not want to learn to operate well. And like, I don't even judge people for that necessarily? I love them, to be clear, but that's something that's true for me and not necessarily everyone else, and if you don't care about something, you're not going to put your best effort into it. It becomes a chore. A necessary step to doing what you actually want to do, and as a result, people don't try and they just suck at it.
Edit: I'd much prefer and think it's more practical to create cities where cars are an option, not a requirement. If you WANT one, you're welcome to have one. And maybe it drives itself, or maybe you drive it. With those options being on the table, we can then raise the bar for driving tests which we badly need to do, there needs to be a higher skill floor for driving and more stringent requirements, especially as people age. Maybe once you get to a certain point in your life, as most of us probably will, you just can't have a manual car anymore. Then you can buy one that drives itself, or just join people on a robust public transit system.
This doesn't need to be hard, but it's made hard because tons of entities involved have a financial stake in keeping us dependent on cars. I think that's why self-driving cars are getting so much traction and funding, because then they get to sell you a new, more expensive car, instead of actually solving the problems of mass transit. And that just sucks.
Trucks are different because they're just for transportation of goods. As challenging weather arrives they can just wait at the nearest rest stop, pull over on the shoulder, etc.
Your shipping will be delayed, but outside of that I don't think society will care. Totally acceptable as long as weather delays only last as long as a storm does.
That's different from rideshare where people do expect to be able to call a Waymo even if it's snowing lightly.
And you're ignoring the fact that not requiring a driver is significantly cheaper. And when it's speed versus cost, cost generally wins. So yes, society is absolutely going to be fine with trucks pausing in bad weather, because it means cheaper shipping.
(And stuff that is urgent and requires next-day delivery is sent across the country by plane anyways, not by truck.)
We can either legislate that or they need an effective union.
Trucks? waiting? In "just in time" culture of storage management, where buffers are low and delays are nightmare?
But I think the bar for civilian is also woefully out of date now. Giant trucks should require a commercial license. Make it easier to get than a delivery truck license, but weed out the soccer moms and bring back the station wagon. If you’re a professional tree trimmer, general contractor, or a forester, renewing your license is on the clock and not a big deal.
First of all for vehicles with total mass of above 3500 kg you need extra license. And also separate one if you have a trailer. This is get it and keep it, until certain age. But good enough often.
And then in general commercial operations also need more licensing. Which needs to be renewed after certain time.
Truth is, I'd prefer the bar at 4,000 lbs, which would limit us to Camrys and CRVs, but 5,000 would really allow just about any reasonable vehicle.
And big heavy EVs with massive acceleration are just too powerful for somebody who's got a Starbucks in one hand, a cell phone in the other, their knee on the wheel, and shouting kids in the back seat. "Pedal misapplication" will go tragic really fast.
5,000 lb limit.
Sincerely,
A Pedestrian
I would be for a CDL for any vehicle that is tall enough that an aver age pedestrian standing on the street cannot see over the hood.
I'm over 6 feet tall and I regularly have trouble doing this as a pedestrian in the USA. I can only imagine how it is to be smaller.
One of the heuristics built into us (because we're mortal beings living beings in a competitive, historically resource-poor environment) is that we trust the devil we know more than the devil we don't, and so unless there's a strongly compelling reason to trust autonomous driving devices a lot more than humans, there will be some inertia against using them, even if the calculations are that it will save X number of lives. I mean, inherent in that calculation is a level of uncertainty and people don't necessarily trust that number, because they don't have a reason to trust it, because they haven't really seen enough to trust it. Why take a company's word for it that it's safer when they have a financial incentive to do some creative stuff to get their marketing pitch? I would say that if you feel it doesn't reflect well on us, it's because it hasn't been thought about enough.
An autonomous vehicle has an uncountable number of operating modes, and it is not feasible (perhaps not even possible) to test it in all possible conditions and states. Even if you could, doing so for (say) every single software change would take years each time.
Maybe that does mean that this is a fool's errand, and we just shouldn't be building autonomous cars, at least not until we have AGI that can think and act faster and with better judgment than a human.
I personally do think that "better record than a human driver" should be sufficient (perhaps with some significant, TBD margin; 0.1% better is probably not enough), but I accept and agree with your toplevel comment that sort of thing won't fly in the real world. The bar is really more like the self-driving car has to avoid making the specific kinds of mistakes and illegal/unsafe acts that a human driver would do (all while not creating new classes of mistakes that a human driver would not make), and, on top of that, be better than a human driver in situations where crash would not be deemed that human driver's fault.
Half the taxi and Uber drivers in New York are perpetually on a phone call, and frequently interacting with their phones.
I suspect — no, I hope — that anyone who admits in advance that they intend to break the rules, won't get a license.
If an autonomous system can get a CDL, it's probably gonna be more effective at continuously meeting that standard than a lot of the humans that do the same thing (but are distracted on a given day or have started using substances, or didn't sleep well or whatever).
That is not sufficient for self-driving vehicles. The license is the thing that shows "okay, this person seems safe" but then some of the things that keep them safe are the threat of accidentally killing themselves, or being arrested and put in jail for a crime they commit while driving, or the financial penalty of being sued for an injury / damage they cause, or the risk of having their license taken away for errors. If a human being applying for their license was invincible, incapable of being jailed, sued, or having their license taken away, you might expect that our CDL or other licensing processes would be more stringent.
It's totally bizarre to give mediocre humans credit for their self preservation and then demand that an autonomous system be beyond superb because it doesn't have it. Statistically, the mediocrity is going to be the bigger problem.
I'll add that one of the penalties for accidents caused by human error is a prison term.
Until a Tesla/Uber/Waymo exec responsible for autonomous driving can serve a prison term for accidents caused by their service, the penalty for accidents caused by autonomous driving is orders of magnitude lighter than those applied to human drivers.
Really? How does anyone get that. I've only seen murder with a car worth 90 days in jail and loss of the car.
In the US, vehicular homicide comes with a sentence that's typically between 9 and 30 years.
But we do allow it. Sure its illegal but we don't put in the effort to actually prevent it. Just because we don't like it and would want it to be better doesn't mean we aren't allowing it.
As we have seen with the titan sub, not even death is a full deterrent. A lot of companies will still risk it for the biscuit and take the risk of getting jailed for their shot at leading the market (in fact, a lot of people risk actual jailtime for financial gain). Having strict regulation and not allowing subpar cars on the street really is the only way.
Note that by good transit system I mean one where you never check the schedule because the next bus/train is never more than 5 minutes (that is a maximum time). By good transit I also mean a system of transfers to express service so you can get to more distance places quickly. Nobody in the US has a good system (NYC's subways don't even come that often!), but there is no reason we can't have that, and some places in the world do have it (at least for a few).
I think you're confusing things. Weather changes midway anyone's drive, and all drivers are required to drive safely and reliably even during sudden extreme meteorological events.
For a road to be closed, it takes an administrative action that reflects a decision that's largely arbitrary. Until a third party makes that decision, any driver is required to drive safely and reliability, regardless of the weather.
It would be interesting to hear what a company like Aurora and their backers were thinking back in 2020. It seems unlikely that they did not come up with these arguments themselves, so what were the counterarguments?
It seems like there are a lot of 'default routes' that trucks take, and in the beginning, it'd probably just be a small number of highway routes that could get clearance in such a way, and then scale up.
Or we could put them on tracks and physically attach them to one another.
Unlike building rail, the concept I outlined would be feasible to do for a private corporation though, even a startup. The delta in necessary resources between the two is huge.
I'm also not quite convinced that you can pack nearly as much cargo on a rail at the same time (with different destinations) as on the road, but the reason for that believe might be the fact that I'm not a rail expert.
I don't really see the point in anything long range for autonomous vehicles in the near-term. We should optimize the use-case of them driving people and goods to and from train stations. For this use-case it should be entirely fine if the max speed they can achieve is 20mph. The stopping times and force that can be achieved at those speeds should bring down the danger by quite a bit.
Growing up we had a neighbor who drove UPS trucks and was rated for double trailers, and that earned him extra money. I gotta assume triples make even more. Doubling the payload per driver mile by making it so any trucker can manage doubles and the old double drivers can manage triples seems doable.
The thing with doubles and triples though is that they are shorter trailers. And it’s probably less likely to pack a triple trailer to the gills than a single, so are they getting 70% as much payload per trailer? If so that puts triples around 2x the practical capacity of a single.
48k miles in the interstate = 80k km, 400k posts, 4 billion dollars (and you can start regionally)
could also run semi-autonomous convoys where a robot truck follows a human truck, extending sensor range
The in cab electronic log system is still largely ignored because states want money from inspections, no autonomous system can handle a tandem adjustment, and everyone wants to put sensors all over roads that barely see replacement let alone investment and dont have maintenance factored in.
Edit: ultimately these companies miss the mark. Autonomous cannot cost more to maintain and operate than traditional fleets and owner operator/independents. It can't force shipping lots to rearchitect the entire parking lot, it can't demand world class networking and sensor systems in jeffrey Wyoming, and it can't wave away traditional safe operation practices with "autonomous"
I don't know how often signs are replaced, but at some point every sign should come with an RFID chip that encodes information about it, although you can probably just glue a chip onto a sign as well.
We scoff at Tesla's self driving claims and the valley between their "FSD" and actual required products, but they'll keep hammering at it. There's probably a trillion dollars of valuation for solving it. Musk's current managerial incompetence aside, Tesla DOES have the data collection platform to get all the data needed for self-driving, especially the "easier" stuff.
Once enough cargo is moving along a route, if there's sections of that route that are not ideal for automated movement, that will be brought to attention of the transportation agency.
The value proposition is obvious: generally, interstate highways are almost totally unutilized from about 10pm to 5am (outside of cities). 6-7 hours of transit that could be used (and offload the transit from rush hour times when passenger cars demand it more).
It involves moving the trailer's rear axles forward or backward based on the load's weight distribution.
Alternatively a drone flying ahead of a truck with cameras could augment the sensing on the truck and extend the sensor range. Deploy the drone(s) from a station on the truck. 2-3 drones per truck for redundancy and charging. Tho maybe drones can not fly at 65mph for long in bad weather.
Driverless trains put about half the system logic outside the vehicles and those things operate mostly incident free
if gov owns the sensors, they no longer have to beg AV operators for crash data
The problem is that they are an absolute nightmare when it comes to highway merging. You really don't want third-parties in-between elements of the convoy, so it essentially becomes a single really long vehicle for merging purposes. On most highways that's simply not a realistic option.
Would require a virtual handshake between the two drivers, but heck why not esp. if you’re guaranteed that the truck behind you won’t try to pass you for an hour :-)
E.g. dedicated lanes for self driving or hand-off stations where human drivers can take over etc etc.
Amid the massive potential upsides, any reasonable government would invest in measures that turn out to effectivley improve self-drivability of the respective national grid.
"Trucking was supposed to be the ideal first application of autonomous driving. Freeways contain predictable, highly structured
driving scenarios.."
Trucking and truck routes do have a strong incentive to switch eventually. But the examples given of Cruise and Waymo are losing tons of money.
Trucking companies would need to rework their business model around these hubs. So they won’t invest in this yet, not until it’s closer.
And for parties trying to make self driving work, it makes sense to try it on smaller, less expensive vehicles first. It’s the number of miles driven that drives learning rates, and if your cost per mile driven is lower, you get more learning per dollar.
And achieving true autonomy also needs a variety of scenarios, not just freeway. So being freeway only isn’t quite the advantage you would want. A million miles of city and highway driving helps you learn more than a million miles of highway driving. I think regulators know this intuitively too, and are more prone to approve.
Anyway, the speed doesn't have to be 50mph. Do 40mph on late night trips. More EV range/gas efficient anyway.
The deceleration would likely be helped by an EV drivetrain where regen and braking can contribute to stopping. Likewise the high-torque EV motors can accelerate a semi far more manageably.
Emergency/Fault stops should likely be handled by convergent infrastructure, or only using routes with sufficient shoulders.
How many scene understandings can be done with on-demand manual takeover? Highways generally have some of the best cell/data networks, and again, convergent evolution of infrastructure.
Sensor distance can probably be improved with the concept of a "scout car". Likely automated trucks would drive in formation (to draft to get better efficiency), a lead car can scout and provide additional effective distance. This car doesn't even need to be a full size car, it could be a smaller teardrop-optimized drone. And if the drone crashes or faults, the following train automatically pulls to the side.
Trucking routes can benefit from neural nets tailored to repetitive navigations of the route. No handling general queries, the tractors (or the control system utilized) isn't driving "a general truck". It is driving from Minneapolis to Chicago.
I think I disagree with the article. It assumes too fast for the vehicle, too general of routes.
Oh, my final disagreement may be the total size of the vehicle. Once automated vehicles can drive at distance ... do you need a massive tractor trailer, or can you instead use more capable (and more cheaply mass produced) vans or smaller form factors? The industry is what it is in current state and a trailer ~= a container, so I guess I see why that wouldn't be a go.
Many cars monitor rotational speed of the tires in order to determine whether a wheel is slipping. A Model 3 can detect whether a tire is new or old based on tread depth, something I learned the hard way.
If trucks were outfitted with rotational encoders on all tires, 'slipperiness' could be monitored periodically by, for example, tapping brakes and checking whether all wheels continue to rotate at the same speed. In emergency braking or maneuvering, the rotational speed of all tires can be input to the recovery algorithm to perform braking. If the brakes are independently controlled, this can be performed per wheel.
More generally, trucks run in fleets and on routes. Similar to aircraft reporting turbulence so the next craft can adjust, trucks can report road conditions so that approaching trucks can adjust their speed or pull over completely if conditions are so bad.
Makes me wonder if it would be more efficient to run trucks as "convoys", maybe consisting of competing companies along predetermined routes. And you could have an ultralight canary car go an appropriate distance along the front to ensure that in the worse case, it is the one that crashes.
Rails are a more efficient interface to wheels, but also more demanding, and less flexible. 10-lane freeways are widespread, but most railways have two pairs of rails at best, and one pair in many places. You can't effectively use them without strict scheduling.
Driving a car in inner Copenhagen is a stressful situation due to the insane number of cyclist you have to watch out for, see https://youtu.be/FaySp9i2zMA?t=113
There's also the "interesting events + training" argument in the article, which I'd love to see points for or against.
Also one of the things autonomy tech could deliver is better driving aids to ease some of that stress (it's not a slam dunk because it is easy to create driving aids that present a false sense of security).
Many examples of high pedestrian density in https://youtu.be/P6sw4EKegp4
The vast majority of humans have some respect for the fact that they are in charge of a ton or more of hulking metal when driving a car, and thus take on more responsibility. Drinking and driving is far less common than drinking and cycling too.
I'd rather take a train where all but the craziest bicyclists venture or let a bus driver handle the consequences of hitting someone. When I'm traveling around the countryside, I take public transit to the edge of the city then rent a car.
They also occasionally run red lights, but less that pedestrians crossing on red.
For example very few countries have "right turn on red" laws like the US has. That law alone is responsible for hundreds of pedestrian deaths. Jaywalking laws were introduced to benefit cars irrespective the fact that they make life insufferable for pedestrians when legal crossings are so few and far between in the US.
And an arbitrary one, in my opinion. We will be racking up millions of driven miles by hundreds of autonomous trucks before that bar is reached. We will have drivers in the trucks. We will have connections to remote operators.
A lot of I80 is really straight, same with other highways. I70 from Denver to St. Louis is incredibly boring all the way through Kansas. I90 from Wisconsin through Minnesota and South Dakota is the same. Just dead straight easy roads. Down south I10 was pretty easy too.
Just because there are a few tough stretches, that doesn't mean the auto-pilot can't drive the other thousand miles on the route, with human support where necessary.
Pick the MVP routes that are easiest. I95 is trivial from Florida, all the way up to DC. Then it remains mostly easy with some really annoying bits around DC, Philly, and NYC. I wouldn't want it trying to get through the traffic over the GW and through the Bronx, but after that it's mostly smooth all the way through Maine.
If anything driverless helps dramatically on this front. If you can run trucks 24/7 without stopping then you can cut their speed say 33% or whatever and still come out on top.
You can also adjust the cargo to reduce stopping distance - use autonomous truckers for things that are large but not heavy.
Or hell even just add more wheels for more stopping traction. Sure gets you more wear & tear but that's an infinitely easier problem than trying to teach an AI to recognise a kid running into the road in a busy urban steet or navigate construction site or whatever other surprises come up.
Author makes some valid points - stopping distance vs sensor range - but to me the conclusion that trucking on highways is hardest doesn't follow at all.
I think that this line of reasoning misses the forest for the trees. If the promise of a silver bullet is a mode of transportation that can safely and reliably work 24/7 day and night with limited to no human interaction and be able to move large volumes of goods, then we already have that: it's called railway.
You probably can’t have cars going 70 or faster sharing the road with trucks going a steady 45.
Well, they should still be driving no slower than the rest of the traffic in the slow lane (modulo what is safe for a larger vehicle under whatever the prevailing road/weather conditions are).
I think a bigger benefit to not needing drivers that need to rest/sleep is that you can allow for self-driving software that can't handle all the weather and road conditions that humans can handle. The self-driving software can pull the truck over for a few hours in a snowstorm that a human driver would be able to handle, and still be ahead of a similar human-driver schedule.
Let's get a little bit crazy and imagine something very different from what we have today, but still only requiring a single driver. One truck cab at the front. Individual trailers have their own wheel motors, their own steering, and a small battery for short range manoeuvring (e.g. at a loading dock/port or). Trailers follow each other at a distance that allows for emergency braking, but close enough that they can be connected by high voltage cables for receiving power from a "battery car". Both the cab and individual trailers constantly communicate, and the moment any participant suspects that all is not right, the whole train tries to slow down and pull over.
If one trailer has a blowout or other issue, it can be left behind to be picked up by another convoy.
At a port or loading dock, the convoy disassembles and individual trailers are driven by remote control, either directly by a human or eventually by the "site AI", so that large articulated vehicles never have to enter the area.
I'm imagining ~12 trailer convoys to balance efficiency gains with disruption to local traffic.
Prepare to have your mind blown: https://en.m.wikipedia.org/wiki/Train
Largely copied from my reply to a similar comment:
Trains requires long-term planning and alignment of a lot more people and organisations, all of whom have their own messy incentives. They're a political quagmire that you might never emerge from.
I'm trying to think of solutions that can be mostly built by the private sector and deployed with only modest legislative changes.
For the same reasons, a friend who works in mass transit talked me around to favouring better "smart bus" infrastructure over lobbying for more rail — at least in most cases where actually getting more rail is an unrealistic outcome.
Just think about it. You're starting a company and the first thing you need to do is spend a few billion dollars before you move your first ton? Doomed company.
This is the classic problem with rail: you need to plan the route so you have sufficient participants willing to fund development, you need to acquire land (which you can't safely do piecemeal because the guy down the line will extort you).
I'm trying to think of solutions that can be mostly built by the private sector and deployed with only modest legislative changes.
I just don’t see a 12 trailer 140m long road train with wires strung between the bits being feasible in any way on a road that has other, untrained users on it.
What problems am I overlooking that are not easily solvable?
So it would be great that as a truck driver, you could attach yourself to one of these trains, sleep for a couple of hundred miles, put the alarm clock, wake up, then detach at the right time and go to your final individual destination. Maybe even take another "road train" for another couple of hundred miles...
People seem to think trucks go from one "transfer hub" to another. That hardly ever happens, and when it does, the trucker usually drops off one trailer and picks up a different one.
Most truck routes are actual deliveries or pickups. You go from the distribution center for a supplier and then you do a chain of stops at 4 or 5 stores where parts of the load are taken off the truck and the rest goes to the next stop. Or you pick up an empty and go to 2 or 3 locations to pick up finished product to either be delivered or brought to distribution.
I mean.. there's _thousands_ of hours of truckers on Youtube just doing their job and discussing it. You could watch any of those channels for a few hours and be _lightyears_ ahead of the typical "AV programmers" thinking on the subject.
It's kind of depressing really, how so many silicon valley type people never take the time to learn how the industry works.
This approach would help data collection (since you have a ground truth from the drivers) and would also mean you have a mechanic on hand to fix the out-of-cab issues that inevitably come up.
Yes, but honestly, you would have the computers look at the traffic report and at the map.
> The boring straight shots they can sleep. Maybe this could get the duty cycle up to 100% and keep the truck moving non-stop? Really any material increase in the duty cycle would be incredibly valuable.
I think technically that's probably possible on many routes, but I wonder how regulations interfere.
They claim that stoping distance and non-normal driving conditions are an issue but these are precisely the same problems that AVs meant for ride share face but with a difference performance profile. The computer doesn’t care if it’s hauling 20 tons or 2 tons, what matters is we give it the correct performance profile and install adequate equipment to operate the vehicle safely.
The only real argument I see here is the limp mode case where it detects an error and must pull off. I can see the actual freeway being a problem here but then we just make sure the routes they initially travel have been checked prior to deployment - it’s a finite road that can be mapped and tested against. This can be further mitigated by having redundant computer systems and redundant environment sensors than it can fall back on to be able to safely pull off the side of the road or even make it to its next stop. Computers can make faster, more-informed decisions about nearly every driving situation and the challenges with operating heavy machinery are well understood.
The reason autonomous trucking has not been developed is not because of R&D, it all has to do with scale and what the funding is chasing. Running a truck is expensive, especially with a large load. Financing a project would have a lot of money dumped into the just the construction or retrofitting of a truck and it’s operation for roughly the same computer as the smaller vehicle - just with a different performance profile. When the vehicle you are testing has a kerb weight of 1100lbs and is electric, your maintenance costs plummet to almost nothing with fuel being electricity. We don’t have the same luxury with trucking. Additionally, Uber, Lyft, et al, are bullish on this technology because they would rather collect fares and pay a fixed cost for vehicles than having to deal with drivers that want silly things like compensation. The gig companies are also looking down a barrel of a loaded, regularly-shotgun that could drastically cut into their revenue. The dollars they are investing into AVs only care about ride share now.
Except in the case of Cruise.
MRC could be handled somewhat successfully for trucks on the highway on sections of highway with good cellular data connection. You could have someone making minimal money play a driving game on the computer all day to keep them paying attention. If an autonomous truck encounters an unusual situation, their screen switches over to a camera feed for the truck, and they can take over the controls of the truck to get it past the obstacle.
Not sure what fraction of highway has sufficient connection to give them a sufficient live feed. Probably a lot though.
This is a weird way to think of it, I don't actually think of it as statistics. The failure cases will be different (between human and AI). Further, we want to be able to sue those responsible (not some nameless company).
Final thought, I wonder if this is the metric, when we'll start seeing insurance remove humans or increase the premiums. Basically, exponentially driving everyone into AI driving
- Sensing distance is a good one for sure. At the same time, it's for now a technical issue. So that a viable temporary progress-making solution might be to have an additional vehicle in the autonomous system: a sensing car is in front, and one or more semi-trucks follow. Then the autonomous system far out-senses a human driver. And we get progress even if the cost-proposition becomes a little harder.
But otherwise:
- Nominal stopping distance as shown include 2.5 seconds of reaction time - fair for a human driver perhaps but hopefully the machine gains a lot of space there.
- "Stopping in lane" is as insane in the city as it is on a freeway. We can't seriously consider that a long term solution in the city? Less risk temporarily okay but not a serious solution. (But fair point on this one: if all else fails it is, for now, an option in the city).
- Even for humans, driving on the freeway is not a question of perfect judgement. It is about quickly but calmly picking a reasonable solution. Including, in doubt, moving to a safer lane and slowing down while measuring the risk of rear-ending. Nobody asks a human to completely eliminate the risk of rear-ending. That sounds within the range of what software can do?
- Jack-knifing and other complex behaviors are probably far better suited to computing solutions - which can sense far more measurements than the humans, independently actuate more control points, and compute actually more or less complete solutions when the human is seat-of-the-pants -ing in real time (with dubious real life record too - see brutal and fatal). Simple no, but all that hard to beat the human sounds absurd (anyone with more input on this?) See for example the drift-parallel-parking stunts achieved with cars. We may not yet be at drift-parallel-parking semi-trucks but anyone here doubts that this is achievable?
- Ease and quantity of samples might be a better point. Waymo really got on it and accumulated a lot of driving data. But then also, most of that data by now is city driving.
- Will trucks happen before cars? Well, cars are already there so... already answered.
Cars are not 'there'.
Air resistance means slower speeds are more efficient (you don't need to worry about human driver needing paid per hour or limited by regulations).
Less other traffic at night, but (presumably) sensors can be designed to work well in these situations.
If you split a truck in 3 it can travel together as a pack for aero yet reduce stopping times as each part brakes seperately.
You could probably even section off an entire lane in certain areas.
Some of these possibly needed EVs to mature more before they become practical.
A train that will take a smaller number of cars ONLY between point A and B. Then at each end offload to a truck.
Not the same as current rail frieght, I am saying a dedicated track between say port of oakland and port of long beach - with some intermediary way points along the way - it just has two lines - there, and back with a dedicated automated train that simply goes back and forth - with gantry ports the train drives beneath at certain locations to lift off cars quickly for that node.
We have done a poor job of leveraging rail in a future-proof manner...
Edited - typo
The trick in driving is not straight lines and clear weather. It’s rainy, snowy, foggy, conditions that make it challenging. Edges cases are the real difficulty, as with everything.
I suppose when your knowledge of the world is limited by a few multiplayer notepad interactions everything is simple.
The reason autonomous trucking is exciting is platooning. You have one truck with a person in it at the head of a long chain of trucks.
This avoids all of the issues mentioned and it's the near-term future of trucking. Unfortunately, this will eliminate many long-haul trucking jobs.
It seems the basic argument is that sensor range is not always long enough to fully encompass stopping distance, but the data presented in the article shows that this is only the case at 70mph.
At 50mph, it seems that sensor tech is well within requirements in all published conditions.
Perhaps an autonomous truck could simply slow down when grade or surface conditions reduce its stopping distance?
While it is nice to do 70mph all the time, slowing down when you can't see or stop well is exactly what is expected of human drivers.
Another response mentions there are easy routes that could be self-trucked first. That would work.
When I think of "completely insufficient California roads" I think of traffic that's already going a lot slower than 50mph.
There are a few other possible reasons; such as electrical trucks still lagging a bit behind electrical cars in terms of volume production and a few other things. Electrical trucks are a lot easier to drive for humans because they don't have a lot of complicated gears, hydraulics, and other things to worry about. That makes them easier to control autonomously as well.
Other than being bigger, the job of autonomously controlling a big truck isn't that different from that of controlling a car. If you can do one, you should be able to do the other. Ballpark it's the same kind of problem. Same roads, obstacles, problems, traffic situations, challenges, etc. Same everything, basically. I don't think the speed or the size of the vehicles is that much of a factor for how hard this is to do.
As for safety, the barrier is actually pretty low. There os a high number of accidents that involve truck drivers asleep or unfocused behind the wheel because they've been driving for way too long. Or getting distracted by their phones. Happens more in the US than in Europe because some differences in rules related to mandatory breaks. But it happens on both sides and it's a bit of a challenge.
We're measuring with very different standards for humans and autonomous vehicles. With autonomous driving it's all about hypothetical things that may or may not go wrong and therefore we should wait until it's so perfect that it can never happen, 100% guaranteed. With human drivers, it's just an endless stream of never ending accidents, fatalities, and misery where we just go "ah well, that's just life". Never mind all the obvious issues with aging truck drivers that are severely out of shape due to life style issues associated with trucking after they've been on the road for ten hours. This attitude is more than a little bit irrational.
I can see that this would allow integration with road trains, as methods evolve.
I saw several posts that spoke about the ground level interaction of bicycle traffic and car traffic being dangerous to bicycle traffic. Years ago we had elevated railways, that came to be unsightly, but functional ways to mix subways(elways) - it made me think that elevated bike-ways might work to reduce bike/car interactions. Bikes are so much lighter in weight that a suspended bike 'mono-road' would work well, it could even be roofed against weather. It has been done in a few places, and seems to work. https://www.google.com/search?q=bicycle+elevated+path&sca_es...
Do you think that no one at UPS, FedEx, Walmart or Amazon has ever looked at how much more efficient or what cost saving they could get if "only they had a rail line to their warehouses verse a road?"
What the discussion is looking at is making a micro improvement to what we have already tested as the better method of delivery, trucks in this case. Where the issue is coming about is we are trying to build autonomous actions into a hybrid roadway verse dedicating a portion of a roadway to autonomous driving. The article states the following:
"Trucking was supposed to be the ideal first application of autonomous driving. Freeways contain predictable, highly structured driving
scenarios"
The term you’re looking for is armchair logistics.
Trains and trucks have been competing for a long time. Autonomous trucks, even in convoys, even where freeways have RFID signs, are vastly different from trains. You don’t need to lay and maintain tracks. You’re not limited to where tracks go. You can assemble and disassemble on demand.
Its a bit ridiculous that in 2024 we have all these apps that tell you about accidents and obstructions but none that actually help re-direct traffic in a meaningful way. You have to hedge your bets on any "suggested" better route.
With smart roads department of transportation could get real analyze on crowded intersections, best place for future on/off ramps, access roads, etc. Traffic congestion and planning is usually an after thought when housing developments go up and then its a catch-up game.
The question is... isn't the problem being approached from an entirely bad premise in the first place? Why would it even make sense to develop autonomous trucks for that part of any freight's journey? Laying train tracks might incur a large initial expense since the tracks, unlike highways, aren't there anyway, but a train is way more energy efficient as only the first locomotive has to deal with air resistance and steel-on-steel is less friction than tires.
I'm talking from a "looking at the wide, societal picture" viewpoint. And from that, it makes so much more sense to shift as much long haul traffic to railways as possible. I'm aware that the mechanics of modern day capitalism are not aligned with what's best for society at large, long-term - mostly that railway infrastructure is orders of magnitude better regarding emissions and energy consumptions, but that it takes many more years for investments to pay off, which is why there is so few railway capacity in the US and what is there, is often enough in shabby condition.
Edit: by the way, why don't we have autonomous freight trains?
Autonomous trucking should start out with its own dedicated part of the highway between distribution centers. We are trying to build autonomous driving to accommodate our current road structure which humans currently drive on. A dedicated / isolated lane with sensors, etc should be the start. Start small and then you can expand after getting feedback data. Everything is always so costly and unattainable because everyone wants to implement it with the current infrastructure but its not needed.
For reference we already do these types of dedicated lanes with ezpass express or HOV lanes.
Is this really correct? What about gatik.ai, they certainly appear to have driverless trucks on the road do they not?
That said the conclusion from the article is really that self driving is hard and we likely will not see it for many years.
It turns out we already have general intelligence (producible in about 16 years) available on demand. And so, the market is already a good indication of which jobs are easy and hard to automate for general intelligence.
Stop the truck using a couple of vehicles, cut open the lock on the rear doors, and loot. Pick an isolated stretch of toad, and the pirates will be gone before anyone can respond.
Even for the people willing to use violence, having a human involved opens up unpredictablity and risk. You can't know what's going on in the head of a random trucker, whether they're armed, whether they fancy themselves a cowboy, whatever. It's a high-risk, moderate reward scenario.
In contrast, if you have a system for consistently hitting autonomous trucks, it's a (relatively) low risk, moderate reward crime.
People deciding to drive while sleepy doesn't mean it isn't scary to do so (in fact it's horrifying). People definitely make illogical decisions. Being afraid while driving isn't even necessarily a logical decision. It's just an emotion we feel. But I know I feel it and a lot of others do too.