Here’s How Knightscope’s Security Robots Surveil the Public

OneZero obtained a presentation that reveals how Knightscope uses facial recognition and license plate readers to track individuals

Left: Surveillance footage captured by a Knightscope robot. Right: Knightscope robot.

Knightscope, the company behind the robot that famously killed itself by falling into a Washington, D.C. office center pond, sells robots to automate surveillance. Since 2015, the company has rented out its security robots to police and private institutions like hospitals and casinos.

These robots, which range from stationary kiosks to roving bulbous obelisks on wheels, collect immense amounts of data using tools such as facial recognition, automatic license plate readers, and wireless device detection. OneZero has obtained a previously unreported Knightscope presentation that shows just how much data Knightscope robots can collect.

The slides—presented in full at the bottom of this piece—were presented to the city council of Huntington Park, CA, in June 2019. The city signed a $240,000 contract to lease a Knightscope robot for three years in November 2018, according to documents obtained through a public records request. Emails from the Huntington Park police department detail the minutiae of employing Knightscope robot, like an extra $1,000 to $2,000 charge to transport the robot to and from a city council meeting, as well as measuring elevators to ensure that the robot could fit through the doors. Huntington Park is one of three police agencies who have contracts with Knightscope, according to the Knightscope website, alongside Hayward, CA, and the on-campus police of North Central Texas College.

The slides also detail the software used by police to control the Knightscope robot, and how the company analyzes the collected data. Knightscope did not immediately respond to a request for comment.

Slides from a Knightscope presentation made to Huntington Park, CA, city council. June 2019.

One presentation slide features the company’s facial recognition capabilities, indicating that the company can surface a known person’s name, the similarity of the person’s face compared to a known image, and a log of other identities that the robot…