Self-driving semi-trucks are coming to America's highways | Hacker News Reader

Self-driving semi-trucks are coming to America's highways

washingtonpost.com

25 pointsbookofjoe56 comments

There seems to be a lot of skepticism in this thread, which I understand, but I really think many people here are underestimating just how much economic incentive there is to make this happen. The freight companies want this, the insurance companies want this, and the truck manufacturers want this. This is not an if, this is a when.

Until they start causing a bunch of problems like the Cruise taxis

That doesn't change the desires.

A certain amount of problems are a fact of life. Do you have a particular reason to think the amount will be overwhelming here?

Tesla has been doing self driving tech for far longer than anyone else in the space, and they're not near perfect. But really it comes down to public perception right? I think it would only take a couple of big issues to make the public distrust the tech. If Cruise had issues with small taxis, a massive semi is going to be a lot more scary and cause bigger problems.

The main thing is that self driving semis are fine if it's going through Nevada or death valley or some other sparsely driven road that doesn't have any intersections. But once the semi enters a suburb or big city...it's going to be an issue.

Doing the same thing over and over longer than anyone else isn’t progress.

No they haven't. Google started in what, 2009?

And Tesla just seems rather unserious. It's one thing to miss a prediction several years out, it's quite another to say you're confident "this year" or "next year" with specifics, repeatedly, and constantly miss by miles. Not to mention the obsession with having fewer sensors. Teslas with the FSD beta still routinely screw up basic driving tasks, not edge cases, like deciding to try and run into a cement wall straight ahead.

I’ve been using Tesla FSD happily for almost a year now

Cool? But we're talking about L4 systems, not L2 systems that can't reliably handle even basic scenarios.

I did 220 miles the other day basically without intervention (only intervened a couple times because I was impatient, not because the car was doing something dangerous): FSD version 12 has, to me, basically proven that their model works and it’s just a matter of polish now. I think Tesla is being conservative about the L rating for their system because getting that wrong in a big way would ruin the company.

Personal anecdotal experience in a car != Facts as it pertains to this story.

Semi drivers do that 220 miles (and then some), each and every day, on quiet interstates and havoc strewn city highways.

And we’re also talking big rigs here with significantly longer stopping distances and far less maneuverability.

And they have to deal with the vagaries of weather. A fully automated rig is no good to anyone if the software decides to not proceed with no driver to override it.

My only point is that Tesla’s system is better than people might think from comments like the original one I responded to and has done things I’ve personally experienced that commenters on Hacker News told me were “impossible” to do with a vision-based system (slow from freeway speeds to a stop because of stopped traffic ahead).

The shift in attitude towards Tesla here and elsewhere is really funny to me because it coincides more or less with the time period in which Tesla started to actually deliver the things they had promised years ago.

I'm not saying your experience is impossible. I'm saying that Tesla -- well, Musk -- has been talking for many years about "full self driving" and they still routinely fail at basic driving tasks. Sure, they probably succeed 99% of the time at a cement wall, or 99.9%, but that's not high enough for L4.

The fact that after more than a decade of development, Teslas still have trouble identifying normal stationary obstacles in the car's path is a very bad sign. Sure, it's not outright impossible to solve this problem with a purely vision-based system -- but it IS much, much harder compared to fusing with other sensors.

Teslas don’t, as a rule, drive into stationary obstacles :) sure, there are highly publicized incidents like this but, also, there are plenty of single-car accidents where human drivers do this too.

They absolutely do try to, all the time. People just stop them, because again, Full Self-Driving isn't -- it's an L2 system that depends on people saving their Teslas from running into things.

Every time a new FSD beta releases, people post a bunch of videos that include Teslas trying to run into things, and a bunch of Tesla defenders hand wave away every incident as inconsequential, because it's inconvenient to their narrative.

Obviously they're going to upload those videos: the videos that have FSD working as intended don't get the engagement.

> a bunch of Tesla defenders hand wave away every incident as inconsequential

Nailed it!

> Tesla has been doing self driving tech for far longer than anyone else in the space, and they're not near perfect.

I suspect that Tesla made its life much more difficult by betting the self-driving farm on cameras. Which may have been a case of sales team overriding engineering objections.

This is just a guess as I have never worked for Tesla. But I did work on detecting and tracking various things and multi-modal sensing (e.g, camera+radar) is very powerful, as some things that are hard to detect, track or ID using one sensor type will be crystal clear after adding another sensor type.

> I think it would only take a couple of big issues to make the public distrust the tech.

The public, at least in the rich world, has become pretty bad at understanding risks. A single unlikely but highly dramatic event will affect preferences much stronger than a much more likely, but less media-genic thing with the same outcome (e.g., the death of a single person). This is not an easy hurdle to overcome, but it is possible if self-driving companies put more energy into comparing their safety against human drivers in a simple, clean, easy to understand form. My 2c.

Voices that point this out are silenced.

Exactly. The safety issues around autonomous long-haul trucking will be reported on very differently than FSD in consumer autos. Likewise, there may be some human interest pieces here and there on truckers put out of work, but don’t expect a massive public outcry (even though this is going to be a massive economic disruption to rural areas).

https://archive.ph/njHKH

Huge misconceptions from naysayers in this thread as per usual with anything related to autonomous. Let's establish a few items...

1) Autonomous trucks do not need to, nor will they, drive on "difficult" routes to be impactful. The "Texas triangle" alone accounts for a huge volume of truck traffic. Add the LA to El Paso route on I-10 and you've got another huge volume of trucks. There are similar routes all over the US.

2) The automated part will start as hub-to-hub to make the ODD simpler. Pick up a trailer ~15 miles outside a city, drive it non-stop to another hub ~15 miles outside the destination city, drop it off. Human drivers ferry the loads the rest of the way to/from warehouses, the autonomous trucks can drive around the clock with lower accident rates and push down fleet costs as a result.

3) Just because the trucks aren't ready today doesn't mean they won't be in the near future. Everyone got jaded by the breathless CEOs and pundits a few years ago who were full of shit, but those of us in the trenches always had realistic expectations. Current timelines are looking like 2027 for scale-up on fully driverless trucks. 3 years is nothing in the automotive industry, especially for something as potentially cost-saving as autonomous trucking.

4) Unlike robo-taxis, the autonomous trucking business model is fairly easy to understand, and the economics absolutely can work if ADK costs continue declining.

I'd love to watch a truck put on its' own chains, or deal with any number of situations our truckers deal with on a regular basis. Truth is, self-driving vehicles will only be feasible for the last mile, backing, parking, or fair weather interstate cruising.

https://www.onspot.com/en-US/

> I'd love to watch a truck put on its' own chains

Here you go: https://www.youtube.com/watch?v=aEMT7D7O-ts

Okay, maybe not technically put on, but perhaps you don't actually need a faster horse?

My school bus in a ski town had these back in 2004

Fair weather interstate cruising so that truckers can get some sleep is pretty valuable.

Even if they only solve fair weather interstate cruising, that's the bulk of the time a truck spends on the road.

I imagine a first generation self driving truck would have very low threshold for unsafe driving conditions. Fog, rain, snow, whatever. Easier to wait it out or let a human drive.

Southwestern US has pretty robustly good weather and covers a large percentage of the population. Coincidentally, the biggest port in the country is in LA. Anything that can distribute cargo from there to another hub is feasible.

I do not think it is on anyone's roadmap for there to be, "Ice Road Truckers: AI Edition"

I have some whispers coming across the pond that US Amazon is about to invest bigly in Self Driving AND Hydrogen trucks.

Will make for an interesting POC I think.

Hydrogen infrastructure is a Japanese pipe dream. Maybe it can work for one company, but it's a bridge too far for widespread or rapid transition.

In all the major economies there are significant heavy truck routes and energy demands that are suitable for hydrogen; city maintainance (garbage, etc), routine trade restocking routes, mining (24/7/365 short haul fleet operation that move billions of tonnes), smelter and processing power plants.

These are all being trialed and replaced with hydrogen as we type:

https://smallcaps.com.au/australias-first-hydrogen-fuelled-w...

https://hydrogen-central.com/pepsico-hydrogen-powered-trucks...

https://www.abc.net.au/news/2023-12-29/planned-hydrogen-fuel...

https://fortescue.com/what-we-do/green-energy-research/green...

https://www.riotinto.com/en/news/releases/2023/rio-tinto-and...

Yeah, I've seen how many Hydrogen leaks NASA and Lockheed still get even with decades of experience and sky-is-not-the-limit unit economics.

I'd say, maybe it can work for one country. Japan has and is making huge investments into hydrogen and it's easy to see why. Trading a dependency on foreign oil for a dependency on foreign lithium, doesn't look like a good trade for energy independence. Meanwhile, hydrogen is accessible via electrolysis of water, of which Japan has plenty of access to.

Lithium is also available from the water Japan has plenty of access to.

"By default, driverless passenger vehicles and trucks can ride anywhere in the United States, unless a state explicitly says they can’t. "

how is that possible? what constitutes a "driverless" vehicle? If I build some piece of crap in my garage that barrels down the road and nearly kills ten people, versus one of Musk's cars does the same thing because of a bug in their touchscreen, these are both "shrugs, what can you do?" situations? I'm pretty sure I'd be arrested and Musk would not.

You can find basically exact duplicate of stories like this from multiple years ago. Similar type of claim in the headline, similar type of article that praises some particular company that plans to deploy driverless trucks soon. The company gets some press attention, makes some money from investors, then cashes out, then in a couple more years the company no longer exists and you see another clone of this story about company Y that plans to deploy driverless trucks soon. Repeat ad nasueum. Honestly comes across as a straight up grift. I wonder if the author or publication has any stake in these companies.

The original author likely works for one of these companies, with the listed author trading a little credibility to save a lot of effort an extremely common thing in journalism these days.

If you look around you can often find waves of this kind of article pushed by various PR firms. The sudden influx of articles trying to get people back into the office post COVID being a recent example when an industry realized it was at a major crisis point.

Reminds me of this piece from '05:

https://paulgraham.com/submarine.html

Most assuredly. The thing is the they're always coming until one day they're actually here. Some sort of paradox/axiom

Thing is, when a technology is actually almost here, people stop talking about it. Those on the cusp of making it happen don't want anyone else to know as they don't want to attract competition that might beat them to the punch.

If you're reading articles about how some new tech is "almost here", you can be sure it's not. When radios start to go silent, that is when you might want to take notice.

I don't think you read the same article that I did, because you are impugning Trisha Thadani and/or the Washington Post for grift and conflict of interest with zero evidence. The tone of the article does not at all strike me as praising Aurora Innovation, Kodiak Robotics, or Daimler Trucks. The subtitle warns that they are "speeding ahead of federal regulations," and if you look at her history she has written several articles focusing on the risks of testing autonomous vehicles on public roads.

This concept is like cold fusion. There's talk, there's a company doing this, research doing that, and then nothing. Companies disappear. Cycle repeats.

Until there's a production line, and actual vehicle orders this remains a cyclical news item with no real results behind it.

TBF it's probably more like hot fusion: it'll happen sometime, but probably not in our lifetimes, in light (sic) of the existence proof from the sun and thermonuclear weapons. CF may be impossible.

Waymo has fully driverless rides open to the public in two cities. Progress is slow and incremental, but it's there.

Trucking may be a bit slower because of the liability of a huge truck, but it'll come eventually.

Three cities now, they recently added Los Angeles.

No, LA doesn't have fully driverless rides open to the public yet, as far as I understand it, though it sounds like they're coming soon: https://www.nbcnews.com/tech/innovation/waymo-la-california-...

That should finally put an end to CB radios at least.

Yeah no.

Truckers is a big group of Trump supporters, this will become an election issue.

Well…trucking offers an opportunity (either as drivers or owner/operators) to people in rural areas that otherwise don’t have any great way to make money (read: rural people in formerly prosperous areas).

While it’s not exactly a low skill job - it’s pretty easy to start making serious money (relatively speaking) for someone that can get a CDL.

Obviously these jobs won’t disappear overnight but over-the-road trucking seems like the low hanging fruit for automation and happens to be a lucrative business for the current human drivers.

Trucking is the single biggest career in the US. Threatening that would be political suicide.

what do you mean by "biggest career"?

Probably: participants multiplied by average gross income?

Not the person you’re replying to, but if you add light and heavy trucks it looks like the only comparable career by number of employees is home health care.

https://www.bls.gov/emp/tables/occupations-most-job-growth.h...

That table is showing top occupations by growth, not by total amount, so it could leave out large but not growing occupations.

There are more teachers than truckers (if my quick lazy google is correct) for instance, but it isn’t on that table.

Maybe this is a better one

https://www.bls.gov/cps/cpsaat11b.htm

Reply on news.ycombinator.com