The Knightscope K5 Security Robot That Fell Into a DC Fountain

Facebook ↗WhatsApp ↗

Originally published 2022-06-18. Dates, ages and announcements in this article refer to June 2022.

A Security Robot's Fall Into a Washington Fountain

On 17 July 2017 a Knightscope K5 security robot patrolling an office and shopping complex on the Washington, DC waterfront rolled off its path and into a shallow fountain, where it sat half-submerged until staff fished it out. Photos of the stranded machine went viral, and jokes that the robot had "drowned itself" or committed "suicide" spread across social media. It did nothing of the sort. The K5 has no feelings or intentions; it was an autonomous machine that misjudged the ground it was driving on.

The moment became a small piece of internet history and a handy shorthand for the limits of robots in the real world. Here is what happened, what the K5 is designed to do and why the incident still comes up in discussions about machines in public spaces.

What happened at the fountain

The robot was working at the Washington Harbour complex in Georgetown, a mix of offices, restaurants and a riverside plaza with a fountain. Bilal Farooqui, who worked in one of the buildings, posted a photo of the robot lying in the water with a dry comment about having been promised flying cars. His tweet was shared tens of thousands of times, and colleagues and passers-by added more pictures of workers wading in to recover it.

Knightscope responded with good humour, posting a joke on Twitter in the robot's voice about humans being able to take a dip in hot weather but robots not. It called the fall an isolated event that it was investigating, said no one had been hurt and promised the client a replacement robot at no extra cost. The unit had been on site for just over a week, mapping the property. No official cause was published; the explanation most often suggested was mundane: the robot failed to detect the step or uneven surface at the fountain's edge and drove over it.

What the Knightscope K5 is

Knightscope is a California company, founded in 2013, that builds autonomous security robots and rents them to property owners. The K5 is its outdoor model: a bullet-shaped machine around five feet tall, weighing about 300 pounds (135 kg), that rolls slowly along a programmed patrol route.

It is designed to watch and report rather than to intervene. According to the company, its sensors include:

  • cameras giving a 360-degree view, with video stored for later review;
  • automatic licence-plate recognition, so it can log vehicles in a car park;
  • detection of nearby wireless devices such as phones;
  • thermal imaging and alerts sent to human security staff.

Knightscope has also advertised features it calls forensic capabilities and, in some marketing, gun detection. Those are company claims rather than independently tested results, and the robot's main real-world value is as a visible deterrent and a roaming camera. Its rounded white body led many people to compare it with R2-D2 from Star Wars, or less kindly with a giant egg or a rolling trash can.

Why "robot suicide" was only a joke

The suicide jokes worked because people instinctively read human motives into machines, especially ones that move on their own. Psychologists call this anthropomorphism, and it is the same impulse that makes people apologise to a robot vacuum or give their car a name. The robot even had a nickname, "Steve," short for Security Technology Enhancement Vehicle, which made the story easier to tell.

In reality the K5 navigates with lidar, cameras and pre-mapped routes. Like many autonomous systems at the time, it was good at spotting obstacles in front of it and weaker at reading changes in the ground: kerbs, steps, wet surfaces and water. A shallow, still fountain at floor level is just the kind of feature that can confuse such sensors. The incident was a reminder that real-world environments are messy, and that robots tested in car parks and corridors can still be caught out by a plaza designed for people.

Earlier and later K5 incidents

The fountain fall was not the first time a K5 made headlines. In July 2016 a K5 patrolling the Stanford Shopping Center in Palo Alto, California, knocked over a 16-month-old boy and kept moving, according to his parents, who told local television that it struck his head and ran over his foot. The child suffered minor injuries. Knightscope apologised, called it a freak accident and said it would investigate, and the mall temporarily took its robots off duty. It is not known whether the same unit was involved in both events, and there is no reason to assume so.

Other episodes followed: a man was arrested in April 2017 after allegedly knocking over a K5 in a Mountain View car park, and in late 2017 an animal shelter in San Francisco drew criticism and a city warning after using a Knightscope robot near its building in a way critics said targeted homeless people. Each case fed a wider debate about whether patrolling robots make public spaces safer or simply more watched.

What the fountain incident tells us

Funny as it was, the fall raised fair questions that matter more now that robots and autonomous vehicles are common on streets and in buildings:

  • Safety around people. A machine weighing several hundred pounds needs reliable ways to sense small children, pets and changes in the ground.
  • Privacy. Robots that record faces, number plates and phone signals collect a lot of data about people who never agreed to it.
  • Accountability. When a robot causes harm, it should be clear whether the operator, the property owner or the manufacturer is responsible.
  • Realistic expectations. Marketing often runs ahead of what a robot can do on a wet Monday morning in a busy plaza.

Knightscope has continued to grow, listing its shares on the Nasdaq exchange in 2022, and its robots now patrol car parks, hospitals and campuses across the United States. The fountain photos remain the image many people think of first when they hear the name.

A note on the dates

The fountain incident took place in July 2017; this page was first published in June 2022, when the story was circulating again. It describes a single, five-year-old malfunction, not a new event, and it should not be read as a description of how current Knightscope models perform.

For more on robots slipping out of human control in surprising ways, see our story about a self-learning robot that wandered off its test site.

Sources: Bilal Farooqui on Twitter, 17 July 2017; NPR: when a robot face-plants in a fountain; CBS News: security robot ends up in Washington, DC fountain; The Next Web: the robot's makers laugh off the fountain dive; The Verge: mall security robot knocks down toddler; IFLScience: original report; Knightscope.

COMMENTS