Back of the envelope calculations (3) Amazon’s trails: wings vs knees

Taking as a starting point the previous post on the text about the future of employment by Carl Benedikt Frey and Michael A. Osborne, we can produce some ideas. Home delivery was until now one of those jobs that was deemed as a human task, as the amount of unpredictable situations that could happen made the substitution by machines almost impossible to substitute a person (that anyway earned a limited wage).
Some months ago Amazon, the internet retailer, published some videos about a new ultra-fast delivery system for small loads, Prime air, based on small drones.

The video shows a quite American context: the drone takes off from a logistical base and delivers a small pack by landing on the client’s garden. In a country like Spain, in which most of the people live in apartment buildings, this would have some problems, and the same can be said about other countries in which Amazon operates.
Taking existing drones with a design that seems similar to those shown on the video, as the Parrot AR Drone, the idea of a 30 minute delivery time seems limited: the Parrot cruises at 18 km/h (some 12 miles/h). Let’s assume that Amazon uses twice that speed, so in half an hour you would get as far a 18 km (without considering the time to prepare the pack in the fulfilment centre). This means that if this idea is serious, either Amazon multiplies its fulfilment centres (losing a strategic edge over conventional retailers), or it will limit this system to the areas closest to its 55 fulfilment centres in North America (MWPVL international data for april 2014). The fact is that these centers are in quite peripheral locations, so the population that could be reached would be limited. The following map shows the location of the two Amazon fulfilment centres in the Los Angeles Basin, San Bernardino (open in 2012) and Moreno Valley (to open in 2014), on a heat map rendering of population densities (census 2010), with the road network and a 18 km grid. The idea of delivery on the same day seems much more realistic than 30 minutes delivery, and probably on that time scale the road would be more competitive than a sprawl of fulfilment centres. On the other side, in Madrid the Amazon base is in San Fernando de Henares, just slightly over 18 km from Puerta del Sol, the urban core; in a denser city, a single centre would cover a substantially higher share of the metro population… but as they live mainly in multi-storey buildings drones would have problems.

Delivery by drone raises other issues. Civil aviation regulations are stringent in terms of rules to ensure the safe take-off and landing of aircraft, especially in terms of geometric conditions and electromagnetic interference. It is clearly possible that Amazon could design its fulfilment centres to adapt to those rules, with take-off and landing corridors adapted and without obstacles, which would be easier for choppers. Besides, both LA locations, for instance, are near an airport. However, how do you know if the delivery address complies with such rules? Trees, posts, buildings make a good bunch of potential obstacles. Sure, we have good aerial images, but the problem here is more complex: there would be a need for a good 3D cartography, up to date, with these obstacles, and there is the issue of the liability of a homeowner that by extending his home with a plan-abiding project restricts the air accessibility of a neighbouring property. This would certainly make way more complex volumetric conditions for buildings.
Let’s just go back to the problem: what a deliveryman does today? He comes in a vehicle that the parks (as he can), goes down, and uses the sidewalk to get to the building lot. If the home is individual, he gets to the gate. If it is an apartment building, he must enter a common space, and then use a lift or the stairs. Hard for a chopper. Would Amazon’s idea be closer to Valkyrie, NASA’s robot? At first sight, it seems more feasible, especially in dense cities, but it seems also far away in time. In fact, the most logical solution (even more for denser cities with frequent traffic jams) would be a walking robot… able to run on freeways, passing through cars in a traffic jam and nullifying the parking problem, as it could just go up five storeys of stairs up to any apartment. The issue is to know if this would really cost less (just in economic terms) than to pay a person. This would not touch that much the urban fabric in physical termes, but it would matter in terms of parking… and also in terms of the existing retail basis.