I don't know where you're getting your numbers from, but I'm privy to internal industry estimates for both. The TAM for consumer transportation is vastly larger than commercial logistics, and you can address large/profitable parts of the latter market with solutions designed for the former.
Both markets could probably be made profitable, but their consumer transportation efforts are further along. If you have to cut one, trucking is the better choice.
I'm assuming that consumer transportation also includes train, plane, and boat. The OP talked about ride-hailing/car transportation like cabs and what-not.
Those are two vastly different things. Unless you're saying that self-driving cars/personal vehicles will replace the other things?
Yeah i think they are including personal vehicles. Presumably the tech could be licensed to car manufacturers, I can see a lot of potential there. Not so sure about "we are going to turn all car owners into ride hailers."
Maybe. Or maybe a focus can get something working sooner and then they can expand back. Transport is a larger market, but it may be harder (obvious: transport implies backing the truck/trailer up to the loading dock, rideshare can avoid that - though the above is just a guess)
Waymo doesn't really need investors. And they're arguably at the very bleeding edge of self driving tech (at least in terms of quality). I can't see anyone else dominating the self driving truck market in the meantime.
I was on the founding team of a transportation company. Personal transit is projected to be a much larger industry than what you are calling "transportation" which is actually "freight/commercial logistics", and is already growing faster.
I could believe the union force, but Waymo have a controlled environment in which to carry on testing, namely depots. Alot have their own purpose built shunter's a specially adapted vehicle to pick up trailers and put them on the loading bays, and to take them off. Why do these exist? A lot of truck drivers cant reverse onto a bay to save their life!
The depot will already have copious amounts of cctv in which to enhance the AI driving systems, whilst also being able to keep track of all trailers coming into the yard and leaving via a gatehouse.
So just like warehouses have become automated, to me, it seems like the next step in automation is to handle the trailer movements in the depot. They will still need a human, to wind up or down the wheels, and plug the air in, when moving a trailer on or off the bay, but the warehouse can get busy at peak time, and the demand for a shunter to move trailers off the bay so the warehouse doesnt slow down is a real need.
That's where Phantom Auto pivoted to. They moved away from most on-road products even though they're under development and turned over to driving terminal tractors and forklifts remotely.
Waymo is an US company and public transport is not really a thing in the US, so they'd probably struggle to find funding for that. In Switzerland autonomous buses are being tested already since 2016 on different locations as pilot projects: https://www.astra.admin.ch/astra/en/home/topics/intelligente...
Aren’t robotaxis public transport? They’re a personal rapid transport (PRT) system that uses existing infrastructure (the roads), rather than requiring expensive railways be built throughout every neighborhood.
There’s a light rail project going on near where I live that’s going to have taken near a decade and several billion dollars to complete. “just” scaling up such systems to run throughout suburbs is economically and politically infeasible.
Maybe, but PRT is not mass transit and not useful in general. They might be safer then human controlled cars (this remains to be seen), but they have all the other disadvantages. They still need to park someplace (most are only going to be used during rush hours). They still take up road space for only a couple people meaning a lot of traffic and we need to build a lot of roads. They still pollute (and may be worse if they get programed to drive out to free parking in the suburbs).
The scale of buses make them much worse in terms of cost-effectiveness and therefore environment safety. Instead of spending overhead on specialized equipment, it's much better to mass produce similar typed vehicles that are easier to maintain and service and put that money towards environment measures.
Also, when you factor in things like the carbon cost of the driver that is needed for the bus, the fact that buses still operate on a schedule and not on demand (therefore leading to an entire bus with a capacity for 40 people being occupied by one or two), and that on average they're much more expensive to transition to more eco-friendly advancements, I would say busses are way worse than a fully functional sdc environment.
People envision sdcs as taxis with robots, but it would mean so much more. You could have better and safer bike lanes, slower lanes with with lower speed vehicles that cost ~1k for < 1 mile travel increasing the chances people go and visit nearby businesses. And all of it with the efficiency of a factory instead of the bureaucracy of most transit systems where it takes millions to add the bare minimum of services.
A bus with just 5 people on is already better for the environment than putting those people in cars (1-2 per car), and quickly gets more cost effective. You need a lot less buses than cars to transport the same amount of people.
> Also, when you factor in things like the carbon cost of the driver that is needed for the bus, the fact that buses still operate on a schedule and not on demand (therefore leading to an entire bus with a capacity for 40 people being occupied by one or two), and that on average they're much more expensive to transition to more eco-friendly advancements, I would say busses are way worse than a fully functional sdc environment.
This paragraph is completely false.
There is no reason you cannot have a self driving bus and thus no driver at all.
While a bus on a fixed route will be empty at the end of the route, even in the not very dense suburbs they have a lot more people on it.
A fixed route is an advantage: you can depend on it! You know where it will be at any time and you can plan. Shared flexible routes cannot work - People need to know they will get to where they are going on time and a shared flexible ride must randomly detour to pickup/drop off someone else and then you miss your schedule.
> You could have better and safer bike lanes, slower lanes with with lower speed vehicles that cost ~1k for < 1 mile travel increasing the chances people go and visit nearby businesses
I have no idea what you are saying. People have places to go. They want to get there fast, not slow. People already can bike or walk to close places - that they often drive anyway is proof that speed is important. Your argument is completely in bad faith.
I honestly think, cars in general will need either new Tyre technology, learn to hover or we need to get them off the road in favor of something which doesn't spew microplastics all over the place. It's good we have battery powered cars, but we haven't solved tire and break dust.
It’s being built. Eventually it will be completed and it will have obvious utility. Then people will expand it because of how obviously useful it is. Same thing happened with the Shinkansen.
No one today wants to build railroads, not without being able to make folks like you and me pay for it to an extent that they can make an excess profit. Trains aren't nearly as profitable (if at all) as automobiles because, for the same amount of freight, you are only able to sell to far fewer individual owners and a highly reduced frequency. There are far more opportunities to overcharge the customer if you have thousands of potential renters or buyers, or millions if you can also sell to everyday people.
Plus there's no status to be had from investing in rail, but plenty of it from some hip young people promising trucks of the future today! After all, if Elon is [saying he is] doing it, how can you not believe?
I see 18 wheelers do illegal maneuvers all the time, but as a society we understand that they have no better option most of the time. If I had to encode things like "park in a busy street" and "pull out into the road even though cars are still coming," and document that as company policy/codebase, I think I might get a bit nervous.
The robotaxis struggle with dropoffs and pickups for this reason I think.
In a dense city, there’s often nowhere you can legally stop. Real taxi drivers will often just stop in the middle of the street, or double park temporarily.
Walking home the other day, i saw a taxi which had stopped to pick up on the northeast corner of a crossroads. The passengers were taking their time. A southbound bus was trying to turn off that corner, but couldn't, because of the taxi, so it pulled out into the street and waited. Then the lights changed, and traffic started coming from the east and west; the traffic from the west piled up behind the stopped bus, and a vehicle in the traffic from the east wanted to turn north, and got stopped by the stopped traffic.
The entire junction was filled with immobile vehicles because one taxi had stopped.
In the UK, this is legal. Black cabs can stop wherever they like to pick up and set down. It's an extraordinary privilege.
This is why you have a human driver do the “busy local street” part and then have the truck drive itself for the “follow the speed limit on an almost-featureless highway for 16 hours” part.
We do this for aircraft current state. The pilot manages the autopilot and exception handling. It's sort of silly we don't throw automatic emergency braking and lane keeping in trucks and call it a day.
Self driving companies want to sell us that we don't need a human operator at all. That's what the companies with large fleets of vehicles want so they the expenses of paying humans. Adding an "autopilot" to a truck while still having a human driver in the truck is not saving the company any money. Requiring a human also means that a truck cannot drive 24/7 like if it was 100% computer operated. Airlines deal with this as well by having maximum operating hours just like truck drivers.
So, unless it's 100% computer controlled, there's really not much point in it from the fleet owner's perspective
I've always thought that the fastest to achieve scenario would be: human driver drives truck to the first rest area on the highway, gets out and lets the autopilot do the rest. Vice versa when exiting the highway. You'll need some logistics planning to get the drivers to and from the right places, but that's not so hard; lots of rest areas are connected to the other side of the road (at least in Europe). Is no one working on that?
the thing about airplanes, is that although they are going fast, the margins for exceptions can be a lot wider. In a truck on a highway the margin of error is only feet. I don't know that a driver can really react meaningfully fast, especially if they are zoning out or idle. And, if they're not zoning out but instead paying close attention, what has been gained?
Agree on lane keeping, but my experience with emergency braking has been less than stellar. it seems to mostly react accidentally and occasionally not do what it is supposed to.
You know else I don't see? Railroad tracks going through the entirety of a city, and trains delivering cargo to local stores and other businesses directly.
I don’t know if anyone is arguing for completely eliminating trucks from the supply chain, but there are definitely counterexamples to the idea freight trains cannot penetrate cities at all. I believe there are even some specialized small trains which can deliver goods directly to loading bays of large stores, but I can’t find an example right now.
I question the premise that 18 wheelers should be allowed operate in urban centers at all, let alone be tacitly allowed to resort to illegal maneuvers. It's a convenience and cost-savings for businesses that are able to have a semi truck drive direct to their door, but it creates serious negative externalities for everyone else. The vast majority of cargo headed into the city should be shifted into vans or small trucks that are intended to drive on city streets.
That sounds super expensive and a logistical nightmare. Imagine a pallet that has 1500 products on it, now those are reboxed, take more space, use more fuel because there will be more trips for each shipment. You also are now tracking probably a dozen boxes instead of a single pallet. There are a dozen boxes to open and inspect before accepting the delivery. You will still have the trucks in the streets too because it cant be easily eliminated if you want to keep auto shops or any light industry where the item exceeds the weight two humans can safely carry.
It would make more sense to me to legislate new commercial buildings must include a loading dock. Require developers to make the commercial buildings have shared loading dock so several small bodegas and stores can all share a dock on their building. This would reduce the amount of the trucks parking in the street with a liftgate or forklift to lower the pallets.
There's also a huge space between 18 wheels and Sprinter. Fun fact: There's even cargo bikes that can carry up to 3 euro pallets (though that's clearly stretching the definition of "bike" to the breaking point) https://pulse.dbschenker.com/de/xxl-lastenfahrrad-europalett...
I don't follow this space, so forgive me for the ignorance...
But, wouldn't interstate trucking be one of the easiest self driving applications?
Use a driver to get the truck onto the interstate, pull over to the shoulder, set truck in autonomous mode, get out, truck drives close to destination on interstates only, pulls over to shoulder, human driver gets back in...
Seems like a good stop gap solution...
---
Edit: So, maybe not the shoulder, but even a truck stop or an easy to navigate to area. Basically, removing the complexity of surface streets...
the shoulder of the interstate is not a parking lane. merging back into traffic from the shoulder is one of the riskier maneuvers in driving, and leaving an unprotected human there is even worse. it's okay to pull onto the shoulder from time to time when there's a legitimate need, but it shouldn't be SOP for every trip.
Needing one person per truck waiting at highway off ramps to drive it the rest of the way negates most of the advantages of the system. Unless you can figure out the last mile you aren't making any money switching to autonomous driving, and the last mile problem is much easier to solve with small cars than semis.
But it wouldn't be one person per truck waiting. There would be drivers who just spend all day driving the final 10 miles of the route, from off ramp to distribution center.
Doing an 8 hour shift of that, 5 days per week, would also be a much better quality of life for the drive instead of being a long haul trucker, I assume. This may make it easier to recruit drivers - isn't there a truck driver shortage in America?
I think you need specific parking lots by major cities where you do the man/machine changeover, but otherwise this very much needs to happen.
Aside from not paying a driver, it doubles the utilization of the trucks when they don't have to be parked while the driver sleeps. Huge efficiency boost.
Yeah, people are too focused on the shoulder. Your idea is doing what we do with ships today. Specialize harbor pilots get shuttled out to an incoming ship and then bring it into dock. They do the reverse when the ship leaves port. There no reason something similar couldn't be done for long haul trucking
But I'm not sure how big an opportunity long haul trucking is.
Rail is a good solution to going halfway across the country. Trucking is a good solution for getting from your warehouse to the store ten miles away.
Long haul truckers exist, of course, and I presume make sense in the niches they exist in, but I wouldn't be surprised to find that they're a relatively small percentage of the total trucking market.
Realistically you could build warehouses and so on in locations and in ways that fully support access to highways in automated manner. Worst case have own ramps.
Or just build parking lot in suitable location where trucks are driven to and from.
You would need drivers to cooperate. I believe it is hard finding truck drivers as is. Isn't that why its an industry where there is very high percentage of owner-operators? Imagine you now tell them they'll not even be doing the majority of the driving and 80% of their profit will go to the companies that automate the easiest part of driving. They're not going to like it. So as long as you need human drivers it's not going to happen (unless there is a huge recession and high unemployment, then people will take any job)
One has to imagine the build, test, release iteration cycle for consumer-grade transportation is quicker than commercial trucking. Most trucks have an expected life of 30-40yrs vs 3-10yrs for passenger vehicles (yes, the used auto market exists but luxury buyers always buy the new model). Electrification is a major component in any fleet-based strategy going forward, and that’s going to happen quicker on passenger vehicle platforms than commercial trucks as well.
Also, the potential damage and liability differential between a 90,000lb freight truck and a 4000lb passenger vehicle has to factor in, too. All collisions and fatalities are unfortunate, but better to limit the liability early on in the scaling phase and come back to commercial trucking later once the company and tech are further along.
Anyone who actually has experience with 18 wheelers knew this was never going to work. A lot of my extended family own or operate big rigs. I have been along for a few rides, and it is shocking the amount of focused multi tasking it takes to operate one of those leviathans. I rode for a week with an uncle in in my early 20's, I was lost in life and thought maybe being a truck driver would be a good choice.
In a single week I witnessed, multiple near fatal collision's that had my uncle not been 100% locked in to the road he would have caused. Everything from people texting on the highway at 70+ miles, a person almost crossed 4 lanes of traffic to not miss an exit, and they were applying make up at the same time, people cutting in front him with literal inches away, people don't understand the sheer weight of these vehicles and can't comprehend why they can't break like they do. Funny enough, he preferred the highways vs normal roads. Highway traffic was somewhat predictable even with all of the insane issues, but normal roads ? Try moving an 18 wheeler through a residential neighborhood (yes hn this was the only way) or pull out of a Wallmart, those were those most dangerous maneuvers.
83 comments
[ 4.1 ms ] story [ 151 ms ] threadTransportation is a huge industry compared to just ridehailing (which is perhaps just 1% by value).
And investors care about potential market size.
Both markets could probably be made profitable, but their consumer transportation efforts are further along. If you have to cut one, trucking is the better choice.
Those are two vastly different things. Unless you're saying that self-driving cars/personal vehicles will replace the other things?
I have a toy model for e-trucking + autonomy + middle mile. But its a long hard slog to roll out (ha! pun!)
The depot will already have copious amounts of cctv in which to enhance the AI driving systems, whilst also being able to keep track of all trailers coming into the yard and leaving via a gatehouse.
So just like warehouses have become automated, to me, it seems like the next step in automation is to handle the trailer movements in the depot. They will still need a human, to wind up or down the wheels, and plug the air in, when moving a trailer on or off the bay, but the warehouse can get busy at peak time, and the demand for a shunter to move trailers off the bay so the warehouse doesnt slow down is a real need.
I think that is the point, we should be doing better by the citizens.
There’s a light rail project going on near where I live that’s going to have taken near a decade and several billion dollars to complete. “just” scaling up such systems to run throughout suburbs is economically and politically infeasible.
The scale of buses make them much worse in terms of cost-effectiveness and therefore environment safety. Instead of spending overhead on specialized equipment, it's much better to mass produce similar typed vehicles that are easier to maintain and service and put that money towards environment measures.
Also, when you factor in things like the carbon cost of the driver that is needed for the bus, the fact that buses still operate on a schedule and not on demand (therefore leading to an entire bus with a capacity for 40 people being occupied by one or two), and that on average they're much more expensive to transition to more eco-friendly advancements, I would say busses are way worse than a fully functional sdc environment.
People envision sdcs as taxis with robots, but it would mean so much more. You could have better and safer bike lanes, slower lanes with with lower speed vehicles that cost ~1k for < 1 mile travel increasing the chances people go and visit nearby businesses. And all of it with the efficiency of a factory instead of the bureaucracy of most transit systems where it takes millions to add the bare minimum of services.
> Also, when you factor in things like the carbon cost of the driver that is needed for the bus, the fact that buses still operate on a schedule and not on demand (therefore leading to an entire bus with a capacity for 40 people being occupied by one or two), and that on average they're much more expensive to transition to more eco-friendly advancements, I would say busses are way worse than a fully functional sdc environment.
This paragraph is completely false.
There is no reason you cannot have a self driving bus and thus no driver at all.
While a bus on a fixed route will be empty at the end of the route, even in the not very dense suburbs they have a lot more people on it.
A fixed route is an advantage: you can depend on it! You know where it will be at any time and you can plan. Shared flexible routes cannot work - People need to know they will get to where they are going on time and a shared flexible ride must randomly detour to pickup/drop off someone else and then you miss your schedule.
> You could have better and safer bike lanes, slower lanes with with lower speed vehicles that cost ~1k for < 1 mile travel increasing the chances people go and visit nearby businesses
I have no idea what you are saying. People have places to go. They want to get there fast, not slow. People already can bike or walk to close places - that they often drive anyway is proof that speed is important. Your argument is completely in bad faith.
Don't just take my word for it. https://humantransit.org/2019/08/what-is-microtransit-for.ht... https://humantransit.org/2011/03/how-universal-is-transits-g... is from a real expert in the field that explains why you are wrong starting from fundamentals that do not change.
And, with the weight of EVs, even heavier.
I present the California high speed rail project as Exhibit A, and rest my case.
From https://www.cnbc.com/2023/05/17/why-californias-high-speed-r...
Plus there's no status to be had from investing in rail, but plenty of it from some hip young people promising trucks of the future today! After all, if Elon is [saying he is] doing it, how can you not believe?
In a dense city, there’s often nowhere you can legally stop. Real taxi drivers will often just stop in the middle of the street, or double park temporarily.
The entire junction was filled with immobile vehicles because one taxi had stopped.
In the UK, this is legal. Black cabs can stop wherever they like to pick up and set down. It's an extraordinary privilege.
In the mean time, I expect a lot of lawsuits and weird insurance decisions.
So, unless it's 100% computer controlled, there's really not much point in it from the fleet owner's perspective
FTFY
Agree on lane keeping, but my experience with emergency braking has been less than stellar. it seems to mostly react accidentally and occasionally not do what it is supposed to.
I don’t know if anyone is arguing for completely eliminating trucks from the supply chain, but there are definitely counterexamples to the idea freight trains cannot penetrate cities at all. I believe there are even some specialized small trains which can deliver goods directly to loading bays of large stores, but I can’t find an example right now.
It would make more sense to me to legislate new commercial buildings must include a loading dock. Require developers to make the commercial buildings have shared loading dock so several small bodegas and stores can all share a dock on their building. This would reduce the amount of the trucks parking in the street with a liftgate or forklift to lower the pallets.
The reality is that 18 wheeler s are there for a reason, and probably you should know it before think urban center only needs small trucks.
But, wouldn't interstate trucking be one of the easiest self driving applications?
Use a driver to get the truck onto the interstate, pull over to the shoulder, set truck in autonomous mode, get out, truck drives close to destination on interstates only, pulls over to shoulder, human driver gets back in...
Seems like a good stop gap solution...
---
Edit: So, maybe not the shoulder, but even a truck stop or an easy to navigate to area. Basically, removing the complexity of surface streets...
The shoulder has many uses, depending on where the road is located.
In Chicagoland, Pace transit buses use the shoulder, and have giant "Authorized to use the shoulder" signs on them.
In the southeast, driving in the shoulder lanes is encouraged during evacuations, and they're painted with hurricane symbols.
Shoulders can be easily repurposed for other uses.
Never assume that where you are is how things are everywhere.
Doing an 8 hour shift of that, 5 days per week, would also be a much better quality of life for the drive instead of being a long haul trucker, I assume. This may make it easier to recruit drivers - isn't there a truck driver shortage in America?
Aside from not paying a driver, it doubles the utilization of the trucks when they don't have to be parked while the driver sleeps. Huge efficiency boost.
A human driver who handled an outgoing truck could just wait (an hour or so) on the same spot for the next incoming truck.
Rail is a good solution to going halfway across the country. Trucking is a good solution for getting from your warehouse to the store ten miles away.
Long haul truckers exist, of course, and I presume make sense in the niches they exist in, but I wouldn't be surprised to find that they're a relatively small percentage of the total trucking market.
Or just build parking lot in suitable location where trucks are driven to and from.
Also, the potential damage and liability differential between a 90,000lb freight truck and a 4000lb passenger vehicle has to factor in, too. All collisions and fatalities are unfortunate, but better to limit the liability early on in the scaling phase and come back to commercial trucking later once the company and tech are further along.
In a single week I witnessed, multiple near fatal collision's that had my uncle not been 100% locked in to the road he would have caused. Everything from people texting on the highway at 70+ miles, a person almost crossed 4 lanes of traffic to not miss an exit, and they were applying make up at the same time, people cutting in front him with literal inches away, people don't understand the sheer weight of these vehicles and can't comprehend why they can't break like they do. Funny enough, he preferred the highways vs normal roads. Highway traffic was somewhat predictable even with all of the insane issues, but normal roads ? Try moving an 18 wheeler through a residential neighborhood (yes hn this was the only way) or pull out of a Wallmart, those were those most dangerous maneuvers.