Robo grocery delivery?

Tuesday, December 4, 2018
by Cameron Shelley

I was intrigued to read a piece by Joe Dysart in the Communications of the ACM concerning food delivery by self-driving vehicles.  According to the article, there are a number of start-ups working on delivering food parcels to people's doors or curbs using driverless delivery vans.

In brief, people could order food from a service online and retrieve their orders from a vehicle that shows up sometime later with their purchases in a compartment that only their purchaser can open.

Justifications given for these efforts are mainly technical.  With self-driving vehicle technology coming along, it may be possible for an unmanned truck to trundle around a city on its own.  Furthermore, if there is no one in the vehicle, the pilot software could be programmed to favor the safety people in the environment over the integrity of the groceries within.

Of course, it is far from clear that even this advantage will provide adequate protection for public safety.

Be that as it may, it remains unclear why this notion should be a paying proposition.  It has been possible for many years to hire a human driver to deliver food, presumably at a modest cost.  Certainly, our streets are full of courier drivers handing out parcels hither and yon.  Yet, food delivery remains notable by its absence.

There are two exceptions.  The first is the well-known pizza delivery service.  Pizza delivery turns a profit for a number of reasons: Delivering only one product makes for efficiencies of production.  Pizzas are the Model T Ford of foods, that is, produced uniformly in large quantities.  Also, pizza delivery drivers are paid very poorly, an arrangement favored by the industry itself.  In addition, pizza remains a popular staple and so is ordered in large quantities.

The second exception may be milk delivery.  Household delivery of milk was once common, peaking sometime around 1960.  However, it declined rapidly with increases in car ownership and the advent of large, suburban grocery stores offering big parking lots and low prices.  Today, home milk delivery is enjoying something of a resurgence with more interest in local foods and increasing penetration of the Internet into modern households.  Also, like pizza, it is a very popular item that can be the focus of a whole production and delivery system.

Which brings us back to automated grocery delivery.  Unlike pizza and milk, grocery delivery involves moving a large variety of goods in zillions of combinations.  I can only think this might pay off if the popular items, such as bread, milk, and bananas, are used to subsidize others.  

Also unlike pizza and milk, these companies are betting the farm on driverless vehicles.  Why not just hire drivers, who seem to be available at fairly low wages?  

One possibility is that food delivery requires larger vehicles in order to be profitable.  A driver who delivers food in their own, personal sedan might be able to deliver only two or three loads at a time.  Such a load factor might not be sufficient economically.  Unlike ride hailing services, grocery delivery companies cannot offload the cost of vehicles onto their drivers.

Another possibility is that grocery delivery companies do not believe they can keep wages low enough.  Pizza companies use their scale to keep driver wages low and reliant on tips.  Since grocery delivery services are all small start-ups, they cannot be confident that they can achieve this sort of position.

Lastly, grocery delivery services may be relying on rapidly increasing the number of delivery trips.  A typical suburban family in North America today visits their grocery store about 1.6 times per week.  People are inhibited from making more trips by the cost of time spent in traffic and the need to perform other tasks.  If these costs can be reduced by home delivery, then people may order groceries more often.  More trips may make such businesses more profitable. 

Automated driving may make some sense in this scenario, as delivery trucks ply similar routes many times a week, and need little new training and no employee retention efforts.  

Of course, there may be an element of technological determinism in play.  That is, grocery delivery requires cars, driverless cars are the inevitable future, so grocery delivery with driverless cars is the inevitable future.  This sort of thinking may make it easier to get money from speculative investors.

Of course, people involved in technology studies, like me, tend to be skeptical of this sort of thinking.  Still, if I am hauling my groceries out of a Johnny Cab in five years time, then perhaps we will see who's skeptical then.