Starship Technologies has taken its delivery robots off the streets of Sheffield five months after putting them there. The low machines, built to carry supermarket groceries and cafe orders and deployed in parts of the city in March as a joint project with Uber Eats, are gone. Starship says it has "paused the service" because the trial reached its natural end, and the company went out of its way to add that the vandalism the robots met along the way had nothing to do with the decision.
That last point is the one the company most wants understood, which is itself a signal of how the five months went. Flags — the small markers that help pedestrians and drivers notice a knee-high robot — were repeatedly torn off. Bodies were covered in graffiti. Residents complained that the robots cluttered pavements and got in the way of people walking.
Several robots were vandalized in March, shortly after they were rolled out onto Sheffield's streets
Source: bbc.co.uk
Several robots were wrecked in March, within weeks of the launch. A charging station was damaged. People left traffic cones behind the robots so they could not carry on.
Ryan Case, Starship's head of public affairs and communications, said the Sheffield trial is finished and the service suspended while the company analyzes the results. A small number of vandalism incidents were recorded, he said, which is particularly unwelcome for local businesses that were counting on the robots to deliver their orders. But vandalism was not a factor in the pause, Case said: the company made a commercial decision and closed out the trial phase.
Both statements can be true at once, and the interesting part is what "commercial decision" is doing in that sentence. A fleet that loses flags, collects graffiti and needs a damaged charging station repaired is a fleet with worse unit economics than the one modeled before March. Vandalism does not have to be the reason a trial ends for it to be the reason the numbers stopped working.
Context matters here, and it cuts against reading this as a verdict on the technology. Grocery and prepared-food delivery robots have become ordinary in large British cities over the past few years, Leeds and London among them. In some Leeds neighborhoods they are a permanent fixture of the streetscape. Whatever went wrong in Sheffield was local — particular streets, particular neighborhoods, particular residents — not a machine that cannot work in the United Kingdom. The reactions that follow these robots around are split in a predictable way: some people praise the maneuverability and the convenience, others object that the machines block pavements and carry what they describe as an "alarming number of cameras."
The camera complaint and the pavement complaint are the durable ones, and neither is answered by a trial ending quietly. A delivery robot is a convenience for the person who ordered and an obstacle for the person trying to get past it with a pushchair, and nothing in the structure of a commercial pilot gives that second person a vote. Ripping the flag off a robot is a poor way to register an objection. It is also, in a city where no one asked, the only channel anyone offered.
What the announcement does not contain is any number at all. Not how many robots ran in Sheffield, not how many deliveries they made, not how many were damaged beyond "a small number," not what the repairs cost, and not whether Starship intends to come back. Uber Eats, the other name on the project, says nothing here. A pause that the operator describes as the natural end of a trial normally comes with a finding; this one comes with a promise to analyze the results later.
Starship still runs robots in cities that took to them, so the company can afford to lose Sheffield and describe it as a completed experiment. Sheffield's local businesses, the ones Case says were counting on the deliveries, were the customers in this arrangement rather than the subjects of it, and they now have neither the service nor a date.