Maybe someday it will menacingly order citizens to drop their weapons — or else, a la Robocop.
But for now, the Knightscope crime-fighting robots, demonstrated for the University of Texas at Arlington Police Department and the media on Thursday, are limited to less aggressive tasks like scanning license plates for unauthorized vehicles or alerting authorities to people in restricted areas.
Or, for mall bots, telling shoppers how to get to Macy’s. And they’re always videotaping and collecting data, including markers that help identify nearby wireless devices. Those latter skills helped solve a robbery recently.
But Knightscope co-founder and former UTA student Stacy Stephens has much higher expectations for the technology.
“Long-term, our ultimate goal is to be able to predict and prevent crime,” he said, by analyzing past data with real-time, on-site information collected by the robots. “Then maybe we have the ability to put the robots into a patrol state where they are hitting those hotspots.”
The demonstration model rolling slowly around the halls of UTA’s College Park Center on Thursday morning was the Knightscope K3, a 4-foot-4, 300-pound, bullet-shaped robot, which looked like a slower version of the motorized — and creepy — tackling dummies showing up on more NFL practice fields.
And of course, comparisons with whistling and chirping R2D2, the more charming Star Wars utility droid, are inescapable.
Knightscope, based in Mountain View, Calif., has about two dozen robots employed through leases to about a dozen clients, all in California, including Microsoft, the Westfield Shopping Centers, Juniper Network and the Sacramento Kings.
The demonstration was the company’s first in Texas. The UTA police hosted the event and “definitely have an interest” in the product, though no decisions have been made, said Assistant Chief Patrick Bridges.
“We’re always looking for new ideas and new technology that will make our jobs safer and easier and protect our communities,” Bridges said. “It’s amazing technology.”
The robots roll around at up to 3 mph within the confines of a programmed “geo-fence” that defines their beats.
The robots detect crimes or violations using a video camera and an array of sensors, including thermal imaging and air quality sensors, and a microphone that someone in distress can use to talk directly with security.
Officers can use smartphones to sync with a robot for real-time viewing of video and alerts that could help them do their job better. For example, the robot could detect a black-listed license plate number of a dangerous person, Stephens said, providing a chance to “keep domestic violence from becoming workplace violence.”
The K3’s sibling is the K5, a taller, more durable version for outdoor patrols, and a four-wheel version for scaling rougher terrain will be launched later this year, said Stephens.
The robots can’t be purchased, and Stephens wouldn’t divulge the cost to build one. They’re available only through leases of one, two or three years, at the equivalent cost of paying a robot $7.40 per hour. For maintaining that control, Knightscope provides maintenance, repairs and software upgrades for all of its robots.
Stephens attended UTA from 1990 to 1993, pursuing an aerospace engineering degree, but he dropped out after he was seriously injured in an accident. He got into law enforcement in 1998, including a stint as a Coppell police officer.
He also started “dipping into the entrepreneurial world and starting my own companies,” he said. “This is now my fourth company, and all have been public safety-related.”
He co-founded Knightscope in 2013, putting its first robot on the market in May 2015.
The robots aren’t intended to replace any humans, but to “augment” the security process by allowing robots to help with the “boring” task of continuous surveillance — “where you just sit behind a desk or go on these walks,” said Stephens. He believes it would reduce the high turnover rate — and expensive retraining — in the security industry.
“If you take that monotony away from a human being, whose attention span is not that great for those types of things anyway, and you give that to a robot, now you have better-trained people who can actually use the data,” he said. “They’re going to want to stick around longer because you’re not doing all the boring work.”
Written by Robert Cadwallader for the Star-Telegraph