Robot Delivery: Far-fetched Dream or Soon-to-be Reality?

With the COVID-19 pandemic limiting in-person purchases and close-contact deliveries, delivery robots have been once again brought to the spotlight. In China, autonomous vans made by Unity Drive Innovation first started appearing in cities under strict lockdowns to deliver food for over 2500 total trips. And as COVID hit the U.S., startups like Nuro started deploying their delivery robots. It began testing its R2 robot in April of last year in Mountain View, CA, and soon expanded into a partnership with CVS to deliver in three different zip codes in Texas. While the pandemic has definitely accelerated the adoption of robot delivery, it isn’t difficult to notice that — even in such “optimal” conditions — the scale of robot usage is still limited, which brings the question: when will robot delivery become a widespread reality?

Defining Robot Delivery

To answer this question, we first need to define the scope of the discussion. At face, the term “robot delivery” might seem rather clear cut, but there is more to it than meets the eye. Through years of supply-chain innovation, delivery nowadays is a complicated process that involves shipping, labeling, packaging as well as transportation between several different fulfillment and distribution centers. Among these different steps, almost all have been successfully automated — except the last mile, the actual delivery to your doorstep. Until that stubborn last mile is also handled by robots, “robot delivery” in the colloquial sense is never a reality, so that will also be the focus of our discussion.

The Costly Last Mile

From the times of the pony express till now, we have invented transatlantic shipping, complex supply chain management softwares and even warehouse management robots, but one thing never changed — a courier standing in front of your door and handing you that package.

As everything else in the delivery process becomes more cost efficient, that stubborn last mile is making up a higher and higher percentage of the cost structure. It is now estimated that over half of the overall cost of delivery comes from just one mile. This is why companies like Amazon have spent millions of dollars since 2013 on delivery drones and robots, and why the question being posed holds such economic significance.

Challenges of Robot Delivery

So with billions behind the idea, why hasn’t robot delivery become a widespread reality? The COVID experiments have taught us a few things:

1. Environmental challenges: Perhaps the biggest challenge to delivery robots are complex environments. There is a reason that Nuro first tested its R2 robots in Mountain View, a small city with mostly suburban layouts, little traffic, open side walks and good public safety. In a large city like New York, there would be significantly more traffic and obstacles to navigate through, which poses more challenges for these robots. What’s worse is that some physical limitations simply cannot be overcome even with better algorithms. For smaller robots (dog size) like Starship, they often have no place to travel on a road without sidewalks; and for bigger robots (car size) like Nuro R2, they would have trouble finding places to stop by. As a result, most population centers are off-limits for delivery robots in their current iterations.
2. Regulatory challenges: drones are obviously not subject to most of the environmental limits that robot vehicles face, but waiting in place for them are regulatory ones. Under current Federal Aviation Administration (FAA) regulations, drones cannot fly 400 feet above the ground or directly above people. Once again, such regulations are way easier to obey in a suburban town vs. a large city, where every building is taller than 400 feet and people walk all over the rest of the areas. Delivery robots, ground or air, also have to obtain licenses from each municipality or town they operate in. And as delivery robots become more prevalent, regulatory pressure will also increase due to concerns over safety, equity and employment. These regulations, however, are unlikely to have prohibitive effects on the usage of delivery robots (see the case of regulations against Uber).
3. The Last-Foot Problem: The last challenge faced by delivery robots is a harder version of the last-mile problem. Currently, robots still lack the ability to make deliveries by dropping off packages at the correct places, so a human is always needed on the receiving end. This is why the use cases for delivery robots so far have all been food or produce delivery — things that arrive in short time frames and for which people are willing to wait. To broaden the scope of their usage, robots would need much more advanced object recognition abilities. Given the current advancements made in autonomous vehicles, this next step may not be too far away. However, and you might be seeing a trend here, everything gets much more complicated in an urban environment. A robot may soon have the ability to know which door of a house to drop a package at, but finding the correct location for drop off at apartment complexes often requires complicated decision making, or worse, asking around. These are not capabilities that any delivery robots are coming close to having, making a human-free delivery challenging to accomplish in such environments.

Conclusion

So is robot delivery a far-fetched dream or soon-to-be reality? The answer is, well, “depends”. If you live in a small city or a nice suburban town, you can expect to be picking up food from robots or even getting a package dropped off by a robot very soon. But if you live in New York or Los Angeles, you can only pray that your city invests heavily in infrastructure so that DoorDash may take a chance on experimenting there.