[mortenjorck]

I interviewed a few months ago with one of the leaders in the space, and when I asked the HM my standard “what do you see as the biggest challenge facing the business right now” I got a surprising answer: “the weather.”

My sense is that drone delivery has the same weather dependency as self-driving cars, only greatly magnified.

[onlyrealcuzzo]

Labor ~60% of delivery cost: https://www.dispatchtrack.com/blog/last-mile-delivery-costs-...

If people have the option to get a package ~10% slower for ~60% cheaper, they'll take the slower option the vast majority of the time.

I would have assumed that the savings in labor would be eaten by drones using more energy, but it looks like drones can actually be significantly more energy efficient than light-trucks for delivery: https://engineering.cmu.edu/news-events/news/2022/09/16-last...

Energy is the second biggest input in delivery cost.

[atrus]

Except it wouldn't be 10% slower for 60% cheaper of their item, it's 60% cheaper for the delivery cost which is a small fraction of the overall cost.

[darwinwhy]

Have you looked up small, light items on Amazon and seen that they pack them up packs of large N? They do that to hide the fact that delivery is actually a very large fraction of cost on such items. Food delivery, as mentioned in the OP, is another example.

[Cerium]

Exactly - rule of thumb is that about $4 of any small item on Amazon goes to Amazon. That seven dollar cable is really at best a $3 cable, including costs to get the item to Amazon in the first place.

[emacsen]

This is unintuivite at first, but makes sense. As we approach last mile, economies of scale begin to not only diminish, but invert.

Let me put it another way... it's very inefficient to move cargo by a fleet of cars vs a fleet of trucks, and it's even more efficient to move cargo by train.

But it would be terribly inefficient to move that train to your home. A UPS truck is more efficient than a train, and a guy on a bike might be more efficient than a UPS truck, other than labor.

But if we can get an autonomous vehicle, then the labor goes away, and if the delivery drone is electric, then it can largely be very efficient- going point to point in a straight line.

I hadn't thought about it before, but it makes so much sense.

[emodendroket]

I doubt the cost for the drone is 0% of the cost of having a guy operate a truck.

[emacsen]

Not 0%, but a drone center can have a few people operating many drones, so the labor cost is distributed there too.

[l__l]

Absolutely. In cities which are equipped for it, the small self-driving vehicles make so much more sense, for this reason.

Starship (https://www.starship.xyz/) have been running in a big way regularly in Milton Keynes (UK) for years now; the little robots driving around are a regular feature. Seems to be expanding to Cambridge (UK) too.

I imagine US suburb design makes this less feasible (I think I read that Starship started on college campuses in the states), and that's why we don't hear so much excitement about this approach? It seems to make way more sense to me than drones, mostly for the weather issue

[paulmd]

Does wildfire haze get in the way? Can IR or LIDAR or something see through it reasonably well?

If you built your drone on mostly visual data... or visual-centric sensor fusion... imagine trying to deliver in NYC with The Fog. The tradeoff drones make for their airspace is being at an altitude where there's an awful lot of "terrain".

Fortunately it's not like that here but I'm not in an area that's probably dense enough for drone delivery for a bit (except subdivisions/etc).

[rkagerer]

I'm not surprised at all - that was the every first constraint that came to mind when I first heard about this idea.