Transportation

Starship Delivery Robots Complete One Million Deliveries To Become #2 Autonomous Transport Company


Starship, a US/Estonian company which provides urgent delivery via small self-driving robots on sidewalks, announced today it has completed over one million robotic deliveries (and 1.5 million unmanned miles) to paying customers, a milestone unmatched by any other company and which places them only behind Waymo in the autonomous transportation space when it comes to production operations.

(As an important disclaimer, I was a strategic advisor to Starship during its early growth, and while I am no longer compensated by it, I remain a stockholder. While you should consider this bias in reading this article, I nonetheless believe this milestone is an important one in the annals of robotic transportation worthy of reporting.)

Many teams are tackling the problem of building robotic vehicles to carry passengers and deliver cargo. This takes place at many levels, from automated freight trucks, to transit vehicles, to robotaxis, to small road-based delivery robots and a surprisingly large number of smaller robots which use sidewalks and paths rather than the traffic lanes. This problem is considerably easier — speeds are lower, robots are smaller, and the environment, while quite complex, does not have cars in it. Teams have tackled this problem because it can be solved sooner than the problem of driving on general roads.

Starship was the first such company, founded in 2014, and it and began commercial deliveries to paying customers in 2018. Starship robots are small and light and thus inherently much less dangerous — even if one were to hit something or somebody, the risk of damage and injury is extremely small. Some sidewalk robots are similarly sized, others are much larger and heavier, but all take advantage of the low-speed environment.

Small robots never have a driver in them — it’s impossible — and as such they are the first to have entered no-human-aboard commercial service. Among self-driving road vehicles, only Waymo, AutoX and Nuro have performed deliveries or rides for ordinary paying customers and only Waymo has done it at significant scale, reporting 60,000 miles of passenger-only taxi service near Phoenix, AZ in their report late last year. Waymo has done over 6 million miles of service (most with paying customers) with a safety driver on board but with extremely few necessary safety interventions and no at-fault accidents. Waymo has given over 100,000 rides to paying customers according to a Waymo spokesperson. As such they are generally considered the undisputed number one in the space. Lyft

LYFT
/Aptiv

APTV
/Motional have also given rides (always with a safety driver in Las Vegas but did not respond to a query on the volume.) Nuro pure delivery robots have done deliveries in Arizona but are few in number.

Can this be compared?

The challenge of carrying passengers and driving at speed on public roads is clearly much greater, and so it’s difficult to do a fair comparison. At the same time, the fact that it’s easier is the reason that Starship, Amazon Scout

AMZN
, Uber

UBER
/Postmates Serve, Kiwi and others have developed in that space. (Uber is rumored to be selling off its Postmates Serve business, as it has sold of several side units recently.)

Aimed primarily at urgent and semi-urgent delivery, in many cases a delivery robot will replace a human driven vehicle making the delivery — either the customer driving to the store or restaurant to pick up their take-out or package, or a delivery driver making the run. Each such robot trip takes a car off the road, and replaces it with a robot not on the road, reducing road use, energy consumption, congestion and safety risks on the road. Some trips will have replaced a walking trip or a group delivery (ie. big UPS truck) and thus not have produced this benefit, but a large proportion would indeed provide the gain. It’s even more benefit than might be provided by road-based delivery robots such as the Nuro R2 or those that use a traditional car or van.

Group delivery is efficient, but is only for non-urgent packages, and generally not for restaurant and grocery orders with perishable cargo.

Robotic transportation promises much lower cost as well as safety and low energy use. Food and grocery delivery with delivery drivers has become particularly expensive recently. Customers may not know it, but the various food order/delivery companies like UberEats and DoorDash don’t just charge the relatively small delivery fee, they also take 20-30% of the price of the meal from the restaurant. Many restaurants have ended up just increasing their prices by some or all of that amount, because they can’t afford that much loss of margin. Often, a tip is added because the driver is not getting enough of this margin. In spite of all this, delivery of low cost baskets is not sufficiently profitable. Low cost robots offer the potential to greatly reduce these added costs — and they usually don’t take tips. (You may laugh, but people do sometimes try to tip delivery robots.)

Starship’s initial market was campuses, which feature a population that loves to order out and often doesn’t own cars. They’re also cheap and love the lower cost of robot delivery. The pandemic closed many campuses for a temporary period, but created a boost of demand for no-human delivery in cities.

Starship’s million deliveries also involved over 600,000 hours of fully autonomous operation. While Waymo has not disclosed their count of hours with no-safety driver, at 60,000 miles it’s probably less than 3,000 hours, and even the 6.1M miles of safety driver commercial service at suburban average speeds of 20mph would amount to only 300,000 hours. This is interesting because error rates in transportation are often more correctly measured in hours, not miles. This is even true for humans. Because human drivers have about 1/3rd the accident rate per mile on freeways compared to city streets, it turns out the rate per hour is much more similar than the rate per mile in the different environments.

Some have had criticisms — San Francisco effectively banned such robots over fears of traffic on the sidewalks, but there are many areas with very lightly used sidewalks where this is not an issue. The robots are also well liked by people with disabilities who can’t easily walk or drive to stores and want them to be allowed. Almost all other cities have been very receptive.

Starship reports that while its operation has not been flawless and its robots are always learning, any potential issues with the robot have not resulted in any injuries due to the low speed on the sidewalk. In addition to sidewalks, the robots are also doing 50,000 street crossings per day.

Again, the easier problem helps a sidewalk robot deliver such a result. Sidewalk robots such as Starship’s can stop in 30cm when they encounter a problem. They can then call upon a remote operations center for non-autonomous strategic help. The remote operators do not typically “remote drive” the robots at Waymo, Starship or other companies that use this approach — rather they give strategic directions about where to drive when autonomy is restored. Again, the problem is easier because of the ability to stop, though at the same time sidewalks are actually quite chaotic environments, with pedestrians, pets, scooters, trees, trash-cans, debris and many items not often seen on the roads.

Perhaps among the greatest assets that Waymo, Lyft/Aptiv, Nuro, Starship and others have gained from their real operations for paying customers is that real world experience. To paraphrase Sun Tzu, “No business plan survives first contact with the customer.” Particularly the operations with no human on board, which are the ultimate goal of most players in robotic transportation. (Some trucking operators have interim plans to haul freight with a combination of autonomous and on-board human driving.)

Along with the million-delivery milestone, Starship also announced new operations at UCLA, Bridgewater State U and Modesto CA. They are a frequent sight in places like Downtown Mountain View (home to Waymo.) It also announced the closing of a new $17M funding round, bringing funding to over $100M.

Read/leave comments at this page



READ NEWS SOURCE

Also Read  EU Suggests Date For The End Of Combustion-Powered Cars, SUVs