Driving more intelligence at the edge

To Catch a Thief: How Security Robots Help Make Our Lives Safer

RATE THIS STORY:
Rating Unavailable Rating Unavailable Rating Unavailable Rating Unavailable Rating Unavailable
VOTES SO FAR

Pity the poor security guard. In busy environments such as prisons and crowded shopping malls, they need to be everywhere at once. And in a more isolated setting—on the night shift in a lonely warehouse, for instance—they may struggle to stay engaged and alert.

Security robots don’t have these problems. Instead, they have artificial intelligence that works without getting tired or bored and streaming video to broadcast everything they see, bringing others on the scene no matter where they might be.

Minding the store—or the prison, school, or, even, soon, your home—is undergoing a transformation. Artificial intelligence, streaming video, and other connected technologies are giving us security guards that can work 24 hours a day, can see in the dark, and have eyes—literally—in the backs of their heads.

Using robots for surveillance is a fairly new trend, and an upward one around the world, in warehouses, shopping malls, prisons, office buildings, gas stations and other locations. In the U.S. alone, the market for these machines reached $2.11 billion in 2018 and, according to one report, is expected to reach $3.33 billion by 2024.

“A guard can only watch so many places,” says Thomas Stoker, vice-president of business development at Turing Video, seller of the Nimbo surveillance robot. “These robots extend the range. And the boring work can be taken care of by robots so that guards can perform more critical tasks.

“Technology allows humans to be more human.”

A convergence of technologies

Exploding in popularity, security bots are guarding shopping malls, warehouses, parking lots, hospitals, banks, corporate campuses, airports, at least one gas station, and more. They range in mobility from stationary bots that scan license plates of passing cars to four-wheeled, all-terrain vehicles.

They use a variety of connected technologies to patrol, communicate, document, and even chase down intruders. LIDAR; videography and photography; artificial intelligence and machine learning; simultaneous location and mapping (SLAM); sensors; the Internet of Things; GPS: these digital guards pack everything a security guard needs.

The video cameras are always on and they can run nearly as fast as a speeding bull. One example is Turing Video’s Nimbo robot, its body made by the personal-transportation company Segway, can carry a human guard in hot pursuit for up to two hours, rolling at 12 mph.

Robot guards surpass their human counterparts at certain tasks, which some find worrisome. Will they replace people altogether? Stoker predicts not: he says no matter which security device is being used, whether a security camera or an advanced AI security robot, they are designed to work with humans in a symbiotic relationship; a new tool in the security guards tool belt.

“People are always going to be involved,” Stoker says.

How these robots work

To get a security robot up and running, a human creates the strategy and parameters for where and how the machine will work.

Aided by AI and image recognition software, the robot explores the space to be patrolled. Likely using a revolving light-detection-and-ranging (LIDAR) sensor, the bot maps the perimeter of the room, including all the objects within.

“Technology allows humans to be more human.”
Thomas Stoker
Vice President of Business Development, Turing Video

The human guard then selects points on the map for patrol, the robot traveling from spot to spot in a repetitive pattern while continually scanning the area. Robots can use not only LIDAR and video cameras but also thermal cameras that enable it—and anyone monitoring the robot—to see in the dark.

Throughout its patrols, the robot makes use of its AI and machine learning algorithms to label everything it sees. AI databases and image recognition software is ever growing to recognize more things: animals, vehicles, humans, even specific faces.

“The artificial intelligence knows it’s seeing a person and not a possum,” Stoker says.

A partner against crime

Differences exist among makes, models, and manufacturers, but all these rolling patrollers rely on some type of AI: Should a robot spy an intruder or other anomaly, it sends an alarm to a central security center or to human guards on watch. The responder can instantly see what is happening and remotely confront the suspect. This feature is designed to prevent false alarms—calling the police on an innocent night custodian, for instance.

False alarms can get expensive, Stoker says. Police, fire, and other emergency responders tend to charge fees after a certain number of calls. If the calls keep coming, they may stop responding altogether. When that happens, insurance costs skyrocket.

AI also helps human guards decide how to react to alerts. The AI can provide suggestions and alternatives, making it easy to quickly question an intruder, get a closer look by zooming in, call another guard, activate a siren or alarm, call building operations, alert police, or do something else. After an event, all the data these devices collect becomes available for review and as evidence for law enforcement.

‘The sky is the limit’

As promising as the technology seems, especially for mid-sized businesses unable to afford around-the-clock human guards, robot guards are not yet perfect.

  • Outdoor surveillance can be tricky. For LIDAR to map surveillance areas, walls must be present to bounce the signal back. Without those boundaries, the robot sees outside spaces as infinite.
  • Smaller bots can be hard for cars and trucks to see and may not be appropriate for parking lot patrols. Expensive robotic equipment, as robust as it may be, is no match for a truck with a 4 inch lift kit when the driver isn’t paying attention.
  • Although robots can go up stairs, these designs come with sacrifices that could prohibit their adoption in security, meaning that multi-floor spaces may need multiple robots.
  • Humans tend to regard them with suspicion or even hostility, even attacking them.

In time, these and other limits to robot security will almost certainly be resolved—and new capabilities will emerge, Stoker says. These could include:

  • Calling emergency response, upon request or autonomously.
  • Escorting employees to their vehicles—especially valuable for those who work late at night.
  • Guarding individual homes, especially at night or when the homeowner is away.
  • Detecting leaks, wiring hotspots, and other hazards.

The hidden hand: memory

As robot manufacturers improve the technologies behind bot surveillance, memory and storage—already critical for live-streaming video, artificial intelligence, and other capabilities enabling these rolling patrollers—will continue to grow in importance.

For seamless, continuous video streaming, for instance, 5G cellular technologies will be critical, their expansive bandwidth creating a true information superhighway in comparison with the two-lane road data travels on now.

All that information will need faster processing to avoid “bottlenecks” that can cause pixilation or gaps in recordings. And accessing videos and other large files instantly, when time is of the essence, may call for storing it on the “edge,” close to the machine, either in the device, a removable card or drive, or a local computer.

Many businesses also prefer the confidentiality that edge computing affords, says Taufique Ahmed, Micron Industrial and Multi Market Segment Manager.

“The artificial intelligence knows it’s seeing a person and not a possum."
Thomas Stoker

The video that surveillance robots capture may include an organization’s most secret places—its warehouse inventory; its computer room. Putting this data into the cloud tends to make business leaders nervous: what if it gets hacked?

“There’s still the perception that, once you put data into the cloud, someone could try to use it to their advantage,” Ahmed says. “This is one reason why surveillance systems are putting storage on the edge.”

As a result, SoCs (Systems on a Chip, in which an entire system’s computing components fit on a single chip) are handling more and more images, which take up a large amount of space, he says. Demands on dynamic random access memory (DRAM) and operating storage are rapidly increasing.

Micron MicroSD memory cards can hold vast quantities of data in storage—as much as one terabyte—and our DRAM products, including our top-of-the-line GDDR6, yield data for processing at very fast speeds and in very large amounts at once, accommodating such technologies as 5G and AI.

Designed with edge storage of surveillance video in mind, MicroSD cards include special firmware that enables continuous video recording—critical for security robots’ operation—with minimal frame drops and video loss. And it keeps stored data private by using password lock protection.

A safer world

These emerging technologies enable companies to feel more secure both physically and digitally, and add to the comfort of our everyday lives, as well, Stoker says.

“Knowing a security robot is in your shopping mall, even though you might not even think about it, just makes you feel more at ease,” he says. At home, he says, the same AI acting as your voice assistant by day could be your security guard at night.

“Who knows what the future will be?” Stoker says. “Whatever it is, robots will be a part of it.”

+