Robodelivery and high-end, low-usage equipment rental (and NPR interview)


Earlier on, I identified robot delivery vehicles as one of the steps on the roadmap to robot cars. In fact, these are officially what the DARPA grand challenges really seek, since the military wants robots that can move things through danger zones without putting soldiers at risk.

Deliverbots may well be allowed on the road before fully automated robotaxis for humans because there are fewer safety issues. Deliverbots can go more slowly, as most cargo is not super-urgent. If they go slower, and have a low weight limit, it may be the case that they can't cause much harm if they go astray. Obviously if a deliverbot crashes into an inanimate object, it just cost money and doesn't injure people. The deliverbot might be programmed to be extra-cautious and slow around anything like a person. As such, it might be allowed on the road sooner.

I gave a talk on Robot cars at the BIL conference, and an attendee came up to suggest the deliverbots enable a new type of equipment rental. Because they can bring you rental equipment quickly, cheaply and with no hassle, they make renting vastly more efficient and convenient. People will end up renting things they would never consider renting today. Nowadays you only rent things you really need which are too expensive or bulky to own.

By the way, the new NPR morning show the "Bryant Park Project" decided to interview a pair of speakers, one from TED and one from BIL, so I talked about my robot cars talk. You can listen to the segment or follow links to hear the whole show.

It was suggested even something as simple as a vacuum cleaner could become a rental item. Instead of buying a $200 vacuum to clean your floors once a week, you might well rent a super-high quality $2,000 unit which comes to you with short notice via deliverbot. This would also be how you might treat all sorts of specialized, bulky or expensive tools. Few will keep their own lathe, band saw or laser engraver, but if you can get one in 10 minutes, you would never need to.

(Here in silicon valley, an outfit called Tech Shop offers a shop filled with all the tools and toys builders like, for a membership fee and materials cost. It's great for those who are close to it or want to trek there, but this could be better. This in turn would also let us make better use of the space in our homes, not storing things we don't really need to have.


I didn't quite understand the benefit of deliverbot, as compared to the same service with people. Is it just $x/hr cheaper to leave the person out of the driver's seat, so I'll rent y% more things than I do now? Or is it more about the promise of unusual delivery times that's going to change my behavior?

Certainly the cheapest way to do a vacuum rental business is to hand out the vacuums in-person at a warehouse. Some of your customers drive by your warehouse on their way to and from work, and maybe some of your farther customers would use an existing parcel delivery service. I'm not seeing how deliverybot can change the game, there, except maybe to shave off a little part of the rental cost.

Not own a $200 vacuum cleaner? What a laugh! Whoever
suggested that has never spilled anything, or had a
pet who came inside after rolling in leaves. Why would
I want to wait for a rental in those situations? And
one can get a pretty good vacuum for a couple hundred
dollars these days... what could a $2000 vacuum possibly
do any better?

And if one is renting an expensive, specialized piece of
equipment like a lathe or a band saw, then delivery is
only going to be a fraction of the cost. Robo-delivery
presents little advantage there.

Then there's the problem you probably haven't thought
of -- hijacking! What's to stop criminals to taking
the cargo, or even the whole vehicle to crack open
at leisure?

Consider also the liability issues.

"The deliverbot might be programmed to be extra-cautious
and slow around anything like a person."

Right, because we all know that machinery/software is always
easy to design properly, and always works as designed!

"if a deliverbot crashes into an inanimate object, it just
costs money"

Whoops, there goes all the savings, plus N months' profits!

A blog of crazy ideas indeed... another solution in search
of problem. This is the kind of stuff that makes it into
"in the future, we'll..." articles, but doesn't past muster
when a real-world cost/benefit analysis is done.

While I agree you want your own vacuum for small spills, it would not be up to the fancy one. If the fancy one can be there in 3 minutes, do you even need the small one? 3 minutes, with a 20mph deliverbot just means one unit on call for ever 2-3 square miles (this is the bare minimum, of course, you need more, but it shows what you could get away with.)

It saves a lot more than $x/hour to leave a person out of the driver's seat. To have a driver, you need a truck able to handle a human and the cargo, not the tiny pod for just the cargo. And the driver has to sit there waiting. And if you have 5 requests you need 5 drivers ready at the depot. And you need big depots. A deliverbot depot is a small warehouse where the bot can quickly go down an aisle, grab the desired rental item and scoot out the door. It's a garage somewhere. (After robocars, I expect there will be some garages for rent.)

Handing out in person is actually not necessarily cheaper if it needs a person and a warehouse where people can go and park. And it's vastly more inconvenient for the person. I'm talking about a system where you are at home, and say, "I need an X" and within 3-5 minutes an X is sitting in the garage. To me, if the time is short, it's almost as good as having the thing in my garage all the time. (The deliverbot can be given a code to open the garage door.)

Except in some ways it's better. It doesn't take up space in my garage, it's a much higher quality unit than I would have bought, and it cost me less than buying.

Now a vacuum may not be the perfect example, since you might have to clean it every time which you don't do at home. (Of course, bump up the robotics and then it is cleaned out by other robots, or is a robot itself, like the Roomba's grandson.)

It doesn't matter what it is, as long as it's portable, and a quality unit is expensive. It might be a fancy kitchen appliance or a machine tool. It might be a high-resolution projector for a presentation. It might be a backup disk you want to store offsite or a specialized tool. If it can come by deliverbot for a low price, it could make sense.

I see you ignored the problems of hijacking and liability.
Here's another one to ignore:

Customer: "The robot delivery is here, and the X I
ordered is missing/incomplete/broken."

Service: "Well sir, it was there/complete/functional
when it left."

Customer: "I don't care, it's not now, and I'm not going
to pay!"

Who's right? How to resolve these disputes?

The economics of robot delivery just don't add up. With
big ticket/infrequently-used items, the time & cost of
delivery or picking it up are but a small fraction of
the price, so automation provides little benefit. You'd
want to go the other way -- i.e., flowers or candy, etc.
But there, one hijacking or accident wipes out all your
economic gains.

All I see are lame justifications for some geeks' wet dream
of robot vehicles; a hammer solution reaching desperately
for problem nails that don't exist. You can ignore the
real world, but that doesn't make it go away. Most people
don't want this and there's no advantage.

This is a problem that many systems have to deal with. It is usually handled by insurance and reputation. It can be improved by recordings of the delivery trip, and diagnostics run in the equipment at drop-off and pick-up. But we already handle this problem with rental items today without these fancy diagnostics. Mostly by just eating it.

The hijack problem (piracy) is one that's been talked about for autonomous vehicles. Remember the main driver today is the military wanting to move cargo. They have a few more options available than civilians, but it is indeed considered an issue.

The bots will of course transmit video of the hijacking, and their location, until they are smashed or put in a faraday cage. Hijacking is possible of course but it's not going to be a trivial thing. Quite possibly neighbourhoods where hijackers hang out will not have companies willing to run bots through them if the insurance rates get too high.

Dream or not, we'll see. Certainly people regularly find themselves running to the store for things they need, and would be happy to get a cheap deliverbot service at the very least.

Pretty close to what Webvan was thinking.... They had robotic warehouses, where the orders were packed by robotic systems, then loaded in to the truck. You are taking the next step; the delivery is automated as well. Considering the auto-drive lanes in the San Diego area, I suspect it could be easily done. My original comment to a Webvan exec still remains -- what do you do when there is an extreme slowdown, do you lay-off the robots? (yes, I know, and I said it, but it is true)

Some counterpoints:
a) The addditional transport volume means more energy consumption. It is possible that this is offset because less items need to be manufactured, but given that many items expire with use (i.e. your vaccuum will break down after X hours of use) this is unlikely.
b) You've gotten used to your brand of vaccuum, how to empty it, how ling the cord is etc. Doing Rental means you are going to get a new unit moreoften than if you own one, and it means there'll be a (short) period to become comfortable with that.
c) You still need manpower to check that returned rentals are in good condition or risk frustrated customers.
d) A good location to pilot such a service would be a high-rise apartment building or complex. If this was profitable, why isn't it done today?

While a vacuum was the item suggested to me, I think other items make more sense. Many of these items already are commonly rented, even with the large hassles involved in renting them such as driving to rental stores and doing paperwork.

The deliverbot is going to be a lightweight electric vehicle and thus lower in energy costs. And indeed energy of manufacture (and shipping all the way from China) is going to be much larger in many cases.

I also expect more devices of the future will know when they break, so human inspection will be less necessary. If something arrives broken, just call and another one will be there in 10 minutes. You are the human inspector. If you break it as soon as you get it you might blame the other guy but this will be rare. Insurance covers this.

But the big thing is convenience. Going to the basement of an appartment building for a vacuum is not convenient. A robot at the door or in your garage is.

I'm not sure this will pan out, but many people rail at our consumer culture, where everybody has their own private stuff when the stuff could possibly be easily and efficiently shared with deliverbots. With small light items like DVDs, we already have this with netflix. Perhaps it can happen with more than DVDs.

Add new comment