A few years ago, Eran Shir (who was one of my students at Singularity University and who today has an interesting startup using mobile phones to solve ADAS and self driving problems) suggested that rather than delivery robots, the future might see roving stores. These would be self-driving trucks filled with the most popular items for their region which come to you. You would open them, shop, and automatically be charged for items. From time to time they would travel to a depot for restocking.

Recently a Chinese company showed off a concept mock-up of this idea called Moby, and while the video is a bit silly, the concept opens up some interesting questions without answers.

One one level, sending the whole store, rather than putting your online order into a delivery robot is pretty seriously inefficient -- in energy, cost and use of road capacity. The robots we're working on at Starship Technologies are designed to deliver with vastly less resources. It is better to have a depot where robots can load and have them come to you.

At another level, however, loading robots at a depot is not a solved problem. The Moby approach allows the customer to pick what they want out of the roving store, and the depot can have humans which restock the store who only have to work in brief stints. Delivery robots would need either robotic loading, or full time humans who were always loading, and the latter is not efficient, unles

Another reason people might like the store is for the items they still like to pick by hand, such as produce, and items people like to try on, like clothing. However, while the Moby concept has the shopper buy a pair of shoes, having a lot of clothing inventory would be pretty challenging. Today, something the size of a bus, loaded with stuff, gets around 6 miles/gallon. Operating cost is around 90 cents/mile (plus parking.) There is more if it needs refrigeration. If it's electric it might drop to 60 cents/mile.

Smaller delivery robots with just a few popular items can be quite efficient, especially when they serve multiple customers in an area. Of course, we've had local robots serving us various items for decades -- it's called the vending machine. In Japan, it's perhaps easier to list what you can't get from a vending machine than what you can.

So if the full robotic store is inefficient, is there a public policy solution? I'm usually not up for banning things before they exist, but it will be curious to see where it leads.

There is one thing I do think I might ban before it exists, and that's a robotic version of those trucks that drive around town with just advertising on them. They have no cargo, they are just a billboard on a flatbed. Today they are not that cost effective, but in the future one could imagine the roads becoming clogged with them.

This makes me imagine a general principle -- a robot on the roads should be there in service of a person. That means carrying a person, or going to pick one up, or carrying goods to a person (or company.) A robot that is just cruising around would not qualify. Even with this rule, you obviously need some exceptions which support the human purposes, but don't involve travel to a human:

• Travel to a parking spot to wait
• Travel for maintenance or fuel/energy
• Predictive repositioning to be closer to where need is expected

The last one would allow the store-robot to roam, to some degree, but it would still need to be in the service of people who will use the robot, even if they are not yet identified. An advertising robot would be serving the advertiser, not the people who interact with it. At the same time any rule will cause the immediate finding of loopholes. The billboard would also take passengers if they "happen" to want to circle downtown all day.

But what about something like Google Streetview, which most people find handy? What about robots doing surveillance, or as suggested in the comments below, market research and intelligence gathering (corporate and state?) Will we tolerate data gathering robots only when traffic is light, but forbid them if traffic is heavy? What if it is traffic they wish to study?

There is generally no need for a robotaxi to roam the streets looking for riders the way yellow cabs do. New York yellow cabs spend 38% of their miles cruising around. In fact, other than advertising, I would be interested in comments on other reasons a robot might go on the prowl. Police robots might cruise the streets for surveillance and to show police presence. So, so far:

1. Advertising -- billboards and ice cream trucks, show of force
2. Surveillance -- mapping, data gathering, policing, stalking

Any others?

This is somewhat related to another abuse I imagined almost a decade ago -- the Robo-RV. And not just the RV, but the idea that you could actually have a fleet of RVs which travel with you. When you stop, they would dock together to form a large living area. This idea seems grossly anti-green at first blush, but a simpler form of it actually is efficient: Doing travel in a small vehicle, but when you stop, summoning locally sourced compartments for sleeping, cooking, entertaining or other functions. This approach could actually be better than hauling a whole RV with you on a road trip at 8 miles per gallon.

At least Moby didn't call their concept Bodega.