eevBLAB 102 - Last Mile Autonomous Robot Deliveries WILL FAIL

[EEVblog]

Four reasons why autonomous pizza delivery robots will FAIL.
And nope, they won't save the environment either.
But maybe I'm wrong?
Magna are trialing a pizza delivery robot in Manhatten.

And another "last mile" deluvery boonodggle called Pipedream
https://i.pipedreamlabs.co/
https://twitter.com/eevblog/status/1516598333804265473

[Andrew LB]

I agree completely. You know that they tried that tube delivery system before? It started back in the late 1800's and existed until around WW2. They mostly delivered mail, but some systems delivered food and other larger prototype tubes delivered things as crazy as aquariums and in phillidelpha they sent a sick cat to the vet using a tube. Poor kitty.

[Brumby]

Pneumatic tube systems have been around for ages - but I can't see them scaling up too well.  As for drone systems - well, they will have a number of issues in city environments.

The land based rover idea might develop in time - but I get the feeling it's going to be a tinkerer buried in their basement that's going to come up with something that is cheap, flexible, reliable and efficient ... and even then it's going to get ripped off by some conglomerate.

What we're seeing now are the 10,000 ways not to make a delivery bot.

So, to be clear, while I'm not ruling it out - I'm not holding my breath.

[Black Phoenix]

I agree completely. You know that they tried that tube delivery system before? It started back in the late 1800's and existed until around WW2. They mostly delivered mail, but some systems delivered food and other larger prototype tubes delivered things as crazy as aquariums and in phillidelpha they sent a sick cat to the vet using a tube. Poor kitty. (Attachment Link)

Isn't the tube system still being used in labs and some hospitals?

[nctnico]

I agree completely. You know that they tried that tube delivery system before? It started back in the late 1800's and existed until around WW2. They mostly delivered mail, but some systems delivered food and other larger prototype tubes delivered things as crazy as aquariums and in phillidelpha they sent a sick cat to the vet using a tube. Poor kitty. (Attachment Link)

Isn't the tube system still being used in labs and some hospitals?

Over here you see the tube systems every now and then. It certainly isn't dead.

[golden_labels]

Tube systems are sometimes deployed to pass money from cashier counters in hypermarkets, at least here. No idea if they are still in active use, though. Recently I see workers running with metal boxes and money being delivered from rolls in a pocket.

The problem is that they can’t scale well. As far as I know, major losses come from the working fluid itself (air). The larger the system is, the more energy you put into pushing air around, doing nothing useful. Also a single pipe section can’t transfer more than one container at a time. So the longer the routes, the lower is capacity per unit length.

As for other delivery options, like autonomous vehicles, they have a purpose. Bringing money to whoever sells them . Which I do not take as a negative thing, as long as the offering party honestly believes they are useful and is not blatantly rejecting reality. I see that as an important function in a society.

In general it’s just the good, old finding problems for a solution. A widespread mistake, more commonly seen in going from conclusion to premises.

[manicdoc]

What gets me is the fundamental lack of designing to your environment that goes into these things - small wheels, low clearance, and often a high center of gravity. No wonder they fail. The odd curb, pothole, or step in an unexpected place should not be an issue. Anything bigger the sensor system should pick up and deal with.

Something roughly based on a 4 wheeler one-man off-roader would be a better starting point. but then you need bigger batteries etc. Hay maybe a battery based truck would do it :-)

[Bud]

Yeah... Boston Dynamics should start lending their Atlas robots to pizza delivery shops. That gizmo will solve all problems. It can even run.

[EEVblog]

As for other delivery options, like autonomous vehicles, they have a purpose. Bringing money to whoever sells them . Which I do not take as a negative thing, as long as the offering party honestly believes they are useful and is not blatantly rejecting reality. I see that as an important function in a society.

Everyone in the comments and email is mentioning the Starship Milton Keys delivery robots as the success story and why this works. Sure it works were you don't have rampant crime, and hundreds of miles of dedicated bike paths. No one has shown me the economics of the robots there yet though, and what percentage of people are using it.
Sure you can make this work, it's not rocket science. What is rocket science is getting a solution that works everywhere, and is so vastly better than existing solutions that it becomes a no-brainer.

[Dr Bob]

Flexibility is a huge key issue. The pizza bot looks very tailored to just delivering pizza. How about bread sticks and salad and a 2l bottle of pop?I can think of a whole load of other food examples, but it doesn't stop there. A delivery person might be transporting all sorts of things depending on the time of day and day of the week so anything too specialized is less useful even if it is more efficient for that one thing.

These bots and drones are chock full of very nice parts and might be a lucrative target for a certain brand of collection and re-marketing group (criminals). I expect there is a good pack of Li batteries to be found inside. Will there be seasons of Battlebots where the makers of these things start spotting very familiar parts being used?

These units also don't seem like they'd be that useful outside of a high density downtown environment. It's about 1.5 miles from my house to the local pizza shop (the good one, not the other guys). It IS on the same side of the main boulevard, but no sidewalks and lots of potholes. Just the trip would be something like 15 minutes if nobody messed with the thing and I doubt that would be the case where I live. The government pays a lot to keep many of my "neighbors" breathing and suitably medicated. Besides that, it would be a 30 minute round trip if all went well. That's far too much time to take during the prime pizza ordering hours.

My low carbon solution is to do most of my cooking at home. A big pan of lasagna takes about the same time as a small one and lasts for a good week. I do my big cook a couple of times a week and parcel stuff into containers for freezing/storage. Meals become a pick and choose if I don't want to make anything based on what's on hand. I eat like a king for a pittance and save the planet too!

[EEVblog]

Flexibility is a huge key issue. The pizza bot looks very tailored to just delivering pizza. How about bread sticks and salad and a 2l bottle of pop?I can think of a whole load of other food examples, but it doesn't stop there. A delivery person might be transporting all sorts of things depending on the time of day and day of the week so anything too specialized is less useful even if it is more efficient for that one thing.

All my local pizza shops have icecream/gelato as well.

These units also don't seem like they'd be that useful outside of a high density downtown environment. It's about 1.5 miles from my house to the local pizza shop (the good one, not the other guys). It IS on the same side of the main boulevard, but no sidewalks and lots of potholes. Just the trip would be something like 15 minutes if nobody messed with the thing and I doubt that would be the case where I live.

Yep, same here in the Hills district. It's called the Hills for a reason, and it would take forever to deliver anything anywhere, if it's even physically possible.
Sure you can make these thing work in some areas, everyone is quoting Milton Keys in the UK as if it proves these things will work everywhere, but they won't.

[nctnico]

As for other delivery options, like autonomous vehicles, they have a purpose. Bringing money to whoever sells them . Which I do not take as a negative thing, as long as the offering party honestly believes they are useful and is not blatantly rejecting reality. I see that as an important function in a society.

Everyone in the comments and email is mentioning the Starship Milton Keys delivery robots as the success story and why this works. Sure it works were you don't have rampant crime, and hundreds of miles of dedicated bike paths. No one has shown me the economics of the robots there yet though, and what percentage of people are using it.
Sure you can make this work, it's not rocket science. What is rocket science is getting a solution that works everywhere, and is so vastly better than existing solutions that it becomes a no-brainer.

Why would a solution need to work everywhere? It makes no sense. The best solution is the one that is most economic given the circumstances. These circumstances can vary wildly. For example: the EU is clamping down on delivery services forcing them to pay their couriers properly including all the social & health care premiums. Delivery by drones (whether on wheels or airborne) could turn out to be cheaper at some point.

[james_s]

Around here it would take a few hours before robotic pizza delivery gadgets would start getting vandalized with graffiti and stickers and teenagers would turn robbing them to get free pizza or just getting them stuck into a sport. It really seems like a half baked solution in search of a problem.

[coppice]

Sure you can make these thing work in some areas, everyone is quoting Milton Keys in the UK as if it proves these things will work everywhere, but they won't.

In the videos about those robots in Milton Keynes, when someone opens one, the cavity appears too small to hold much more than a couple of drinks. It doesn't even look like you could get a pizza box into one, which is a key home delivery item.

Milton Keynes was certainly a cherry picked place for the initial trials in the UK. If they can't work there, they can't work anywhere. On the other hand, if they can work there it says very little about their generality.

[coppice]

Around here it would take a few hours before robotic pizza delivery gadgets would start getting vandalized with graffiti and stickers and teenagers would turn robbing them to get free pizza or just getting them stuck into a sport. It really seems like a half baked solution in search of a problem.

That's very much location dependant. There are places where they will rarely get abused. There are others where every time they try to cross a bridge or overpass the robot will be picked up and tossed over the side.

[EEVblog]

Around here it would take a few hours before robotic pizza delivery gadgets would start getting vandalized with graffiti and stickers and teenagers would turn robbing them to get free pizza or just getting them stuck into a sport. It really seems like a half baked solution in search of a problem.

That's very much location dependant. There are places where they will rarely get abused. There are others where every time they try to cross a bridge or overpass the robot will be picked up and tossed over the side.

And therein lies the problem. For this to work and be "game changing" and be worthy of all the glowing articles how it's going to change our lives, it basically has to work everywhere.
This is why I'm skeptical of the Milton keys "success" being used to "prove" that it will work everywhere.

[EEVblog]

Why would a solution need to work everywhere? It makes no sense.

Because the companies behind this are pegging their success on growth and that everyone (or a huge percentage of cities) will want this. They don't want a small niche market, and many may not be able to survive within one.

But ok, let's assume that a company stays small and niche and continues to supply a niche market forever, great. But that means our lives haven't all been changed by these delivery robots like all the articles like to extoll.

[Black Phoenix]

That's very much location dependant. There are places where they will rarely get abused. There are others where every time they try to cross a bridge or overpass the robot will be picked up and tossed over the side.

It already happens with shared bicycles, shared electrical 2 wheeled scooters and other stuff. I don't see why it would be different from this ones to suffer from vandalism and plain stealing/damage.

[nali]

Milton Keynes was certainly a cherry picked place for the initial trials in the UK. If they can't work there, they can't work anywhere. On the other hand, if they can work there it says very little about their generality.

I suspect MK was used because Catapult are based there https://cp.catapult.org.uk/

The UK’s innovation accelerator for cities, transport & place leadership

basically a Gov money-spending outfit.. I've been there a couple of times, the whole place just reeked of self-justification.
They did the driverless "pod" taxis a little while back, whatever happened to them?

[coppice]

This is why I'm skeptical of the Milton keys "success" being used to "prove" that it will work everywhere.

Why does it have to work everywhere? What is suitable for urban environments doesn't necessarily have to be suitable for more rural environments. What it needs is to be broadly useful enough to be a significant player, and build volume. Milton Keynes is fairly small and rather special in its layout. You get modest sized housing estates laid out in a similar way, but a town with 250k inhabitants that is all laid out in this way is rare.

[coppice]

Milton Keynes was certainly a cherry picked place for the initial trials in the UK. If they can't work there, they can't work anywhere. On the other hand, if they can work there it says very little about their generality.

I suspect MK was used because Catapult are based there https://cp.catapult.org.uk/

The UK’s innovation accelerator for cities, transport & place leadership

basically a Gov money-spending outfit.. I've been there a couple of times, the whole place just reeked of self-justification.
They did the driverless "pod" taxis a little while back, whatever happened to them?

Catapult and similar groups are based in Milton Keynes purely because its unique "town planner on crack" design allows things to function which would need to be a lot more sophisticated to work anywhere else. It was used as a test bed town from the day the first phase was complete. e.g it was the world's first fibre to the home test bed, long before the technology was economically viable.

[artag]

The MK trial has expanded slightly, there are 3 shops (co-op groceries) using them near me in Bedford. I think there's another somewhere nearby - Corby, maybe ?

The area covered is standard low-density suburban. A few walkways but most journeys would involve multiple road crossings and normal kerbs. Not town centre or high traffic roads, but not like MK's walking routes either. I've occasionally seen up to 3 waiting to cross a mildly busy junction but not seen any stuck so far.

Here are the locations of two of the stores. The bots are based at these, not at some warehouse.

https://goo.gl/maps/i9bcrWj8undL4Q1EA
https://goo.gl/maps/xxkT6Fht69bjFu3V7

[coppice]

The MK trial has expanded slightly, there are 3 shops (co-op groceries) using them near me in Bedford. I think there's another somewhere nearby - Corby, maybe ?

The area covered is standard low-density suburban. A few walkways but most journeys would involve multiple road crossings and normal kerbs. Not town centre or high traffic roads, but not like MK's walking routes either. I've occasionally seen up to 3 waiting to cross a mildly busy junction but not seen any stuck so far.

Here are the locations of two of the stores. The bots are based at these, not at some warehouse.

https://goo.gl/maps/i9bcrWj8undL4Q1EA
https://goo.gl/maps/xxkT6Fht69bjFu3V7

That looks a lot more typically suburban than MK. I had a quick look around YouTube for videos of people really critiquing how well these things work. They all seem to be people enjoying the pure novelty of them, rather than their usefulness and downsides.

[MadScientist]

We’re 30 years away from truely multi purpose autonomous vehicles. I don’t see any chance if it in my life time.

[artag]

We’re 30 years away from truely multi purpose autonomous vehicles. I don’t see any chance if it in my life time.

I think they'll be a long time too, but many companies are pushing devices with limited autonomy at a level that surprises me. I have no doubt that some niche applications will work. The bots under discussion here are walking-speed devices with small cargo capacity and a low cost of failure, and even then there are doubts as to whether they're economically viable.

But driverless cars with all the capabilities of human-driven cars ? No. 30 years is a long time, but even that could be right.

What might change the game is something like a mandated automated-driving lane on motorways,  protected from random events - essentially a train-mode for long distances. I don't know what might happen to require that, but 4 years ago I would not have expected the level of working-from-home we have suddenly reached. So I'm wary of predicting anything too specific.