Walmart is using fully driverless trucks to ramp up its online grocery business | Hacker News Reader

Walmart is using fully driverless trucks to ramp up its online grocery business

cnbc.com

62 pointskungfudoi26 comments

Interesting. There might be an opportunity there. Instead of product drop offs from a beverage distributor, food distributor, pharmaceutical distributor, etc. and a pickup from a hazardous waste service provider, have the long haul routes terminate at a hub where the truck is loaded with products across multiple suppliers for local stops.

Go to a local store today and you might see multiple deliveries that could be done with a single vehicle in the future. Route planning is not trivial to optimize well, and isn't a core competency for most companies, it's something they need to do.

Maybe Walmart wants to be that logistics supplier. They might have the real estate to build hubs, too.

How do these trucks cope with adverse weather conditions? e.g. sliding down a hill on icy days, dealing with unplowed streets post-snowstorm, driving along where a cop is diverting traffic, or dealing with traffic light outage?

They don't. There is a safety driver on board. The trucks do a 7 mile closed loop. And I bet you money when weather gets bad, the good old fashioned human drives the truck. This is just another fucking non-event hyped by a news website in the realm of autonomous vehicles.

Citation for this?

The second paragraph of TFA says:

Walmart and Silicon Valley start-up Gatik said that, since August, they’ve operated two autonomous box trucks — without a safety driver — on a 7-mile loop daily for 12 hours.

That sure sounds like no safety driver.

Perhaps they simply haven't had to deal with any of those conditions.

Have there been icy roads, snowstorms, construction, or traffic light failures on that route since August?

Registering hypothesis before I read the article: No, they arn't.

They are not yet doing this so at least the tense of the article is wrong. It's likely that there's an internal project but it's nowhere near rollout.

Ok, brb.

...

Post Read: Better than I expected!

The title gave me the impression that the trucks are doing deliveries to end customers. Not the case. They are doing 'middle mile' trips between two known points over and over.

Also they have remote safety drivers watching.

The underlying company is Gatik (https://gatik.ai/), and they carefully map the exact route and add custom logic for it before they let the trucks run free. Seems like a good niche solution!

With two operating trucks I'm still going to call the headline 'mostly false' but I was surprised at both the solution and the progress so yay!

“ Walmart and Silicon Valley start-up Gatik said that, since August, they’ve operated two autonomous box trucks, without a safety driver, on a 7-mile loop daily for 12 hours”

It’s a huge step.

Getting past the tipping point where a technology is used daily in business is important because then there’s a sustainable business. Other companies with imitate and iterate.

Sure, we all want perfection to spring from nothing but that never happens.

Commercial long-haul warehouse-to-warehouse trucking was always the MVP for self-driving.

It has the revenue potential (high utilization), it has the economics (lifetime revenue supports capital investment), and it has the urgency (labor shortage at desired wage level).

Unfortunately, everyone wanted to be the sexy consumer darling.

Yep, I honestly have more hope for this type of autonomous driving. Elon may be right about cameras theoretically containing more information than LIDAR, but that's like saying an encrypted file contains more information than a smaller, unencrypted file. Sure it may be true, but if you can't crack the encryption effectively and repeatably the point is moot.

In the meantime stuff like this is what's going to really make an economic difference. Boring supply chain shit that makes everything marginally cheaper and most people are completely unaware.

Not only that, but it should be pretty clear that using a combination of vision technologies (cameras, LIDAR, RADAR, sonar...) means that more edge cases should be covered. The failure conditions for optical wavelength cameras are different than for LiDAR, or infrared cameras, so really good sensor fusion should theoretically reduce the number of situations in which the internal model of the world has major errors. I'd rather be in a self driving car with 6 cameras and a 360-degree LiDAR then one with 20 cameras, even if 20 cameras are providing more information about the world.

This is analogous to the megapixel quantity versus quality debate. I’ll take fewer megapixels paired with a good lens over more MPs with substandard glass anyday.

Now you just need really good sensor fusion, and a 3rd sensor type to outvote the stalemate of the other 2. Oh wait, now they give 3 different results

I don't have much hope for driverless trucks at least not here in Canada during the winter. A 50 ton 30m long driverless truck fishtailing on ice would be dangerous.

Once you go driverless, the truck size can be reduced since that large overhead (the human driver) that needed amortized over a big load is gone.

They could still join up to make trains at certain points in the journey but even then they'll all be able to brake or avoid obstacles individually.

I don't think there is any intention to shrink the trailer, which represents the bulk of size.

I think Elon is right, but for the wrong reasons. There should be no fundamental reason that you can't drive a car with two really shitty cameras mounted inside the car looking out the windshield and windows. We somehow manage it every day. The amount of environmental information can't really matter that much since we're able to build a good-enough model of the world around our car without any of it.

Why are we setting the bar at human vision? Computers have so many ways of being able to see better than humans. Why would you not want the most visibility possible? Why not be able to see around the car in front of you for more advanced warning? Or why not be able to see around corners? I mean, we put so many cameras in cars now just to give the driver the most visibility possible... and for some reason, we should only use 2 cameras for driving, because that's what humans have? Seems like a low bar to me.

I think you're misunderstanding my point. Whatever breakthrough that enables autonomous driving won't be blindly throwing more data at the problem until it works since it's clearly not necessary for human drivers. Or phrased another way, autonomous driving isn't a data or sensor problem. Humans don't have super advanced sensors or a hive mind that amalgamates all driving scenarios from billions of miles driven and engineers and data scientists just need to catch up. No, it's our brain that's the super advanced thing that we haven't caught up with.

Yes, in real life you would want your autonomous driver to have all the data possible available to it.

Another company doing this is https://www.einride.tech/press

Does anyone know more companies in this space? Some of the legacy truck companies probably do experiments like these as well but their business doesn't depend on it. So mostly looking for startup answers.

I live in Bentonville, but I don't work for Walmart. These trucks are neat, but the online grocery experience is kind of trash.

so it matches the rest of walmart.com you mean

My online order, curbside pickup grocery experience with Walmart has been great, and I intend to keep using it for most things.

Is this the first autonomous vehicle project on public roads to produce actual economic value?

big problem is that they have

1. poor customer service

2. poor online shopping website

3. limited selection ( compared to amazon)

they need to fix this before anything else. I really dont understand why someone would shop on walmart vs amazon.

i agree they are definitely getting better though. i have memberships to both and have been watching how they have improved quite a lot in last few months. the big reason why i got a membership to walmart+ was due to Amazon going from (1 day shipping for last 3 years) to (4+ days!) in the Louisville market, which makes no sense. We have multiple warehouses in the area. Tons of articles/forums discussing it, so definitely im not an edge case. Strangely enough when testing out the membership types (personal vs business) of amazon they are now just finally giving me 1-day shipping on my business account but not personal (verified using both items). Walmart+ has them completely beat on some items that are usually essentials that you can get at the local store. Plus they are offering same-day shipping for free, since they are coming from the store. I've had multiple times where I ordered something from Walmart+ where it said 2-3 day shipping and within hours, it turned into 'Oh we found it local' and delivered it that day.

Interesting how "autonomous" driving is now basically what we do with drones. Remote controlled cars with some driving assistance. How long before this labor gets optimized so only illegal immigrants are willing to do this kind of work? Even better you can outsource this to a poorer national for pennies...

Reply on news.ycombinator.com