Warehouses, factories, and other facilities need to ensure safety and security. In addition to mobile robots, connected sensors and biometric technologies enable operators and managers to use data in real time while also protecting privacy, according to Alcatraz AI.
Robotics and AI can improve the state of security while extending human capabilities, said Vince Gaydarzhiev, founder and president of Alcatraz AI. He previously worked at NVIDIA and Apple and has focused on developing and applying facial-recognition technologies.
“We looked at how to solve security with access control at data centers,” Gaydarzhiev told Automated Warehouse. “Biometrics was not reliable 10 years ago, so the problem was solved with turnstiles, guards, and badging in at doors.”
“We’ve been working on algorithms to create better, faster, and more reliable facial authentication since 2016,” he recalled.
Rock X offers ‘frictionless’ robotic security
“A robot at every door with AI facial recognition can decide on whether to grant access, eliminate tailgating, and help enforce occupancy limits,” said Gaydarzhiev. “Companies choose us to avoid buying turnstiles or hiring more guards.”
Cupertino, Calif.-based Alcatraz AI claimed that its latest offering, Rock X, offers “frictionless” security. It can operate in all lighting conditions and requires minimal human supervision, it said.
The company‘s differentiator is that not only does it control access for large enterprises, but it also provides data, checks for employee consent, and includes a strong back end that can scale to multiple campuses around the world.
“Alcatraz AI is a new category in physical security,” said Blaine Frederick, vice president of product at Alcatraz AI. “Autonomous access control is designed to enhance or replace existing measures. People can use their faces, as with a smartphone, and we make multifactor authentication easy.”
Depending on the customer’s needs, the company’s systems don’t need to store photos and names. It also complies with Europe’s General Data Protection Regulation (GDPR) and U.S. rules such as the Illinois Biometric Information Privacy Act (BIPA), said Frederick.
Source: Alcatraz AI
“A lot of people think maintaining security and privacy is scary, but it’s not if you do it the right way,” he said. “It’s also important to draw a distinction between facial authentication and facial recognition for things like watch lists. We don’t do anything related to surveillance.”
“What comes out of our product are just badge numbers,” Gaydarzhiev said. “All decision making is at the edge, while dealing with enrollments, alerts, visualization, and backups is in the cloud. As we’re deployed, we take data and create analytics.”
The analytics can be integrated with human resources, which is important for making business decisions in real time or predicting potential incidents, he explained.
“For example, if an employee is on a performance plan and goes into a secure area, that could be a red flag,” said Gaydarzhiev. “There’s no other technology to manage that.”
Alcatraz applies AI advances across industries
Alcatraz AI has raised $55 million in the past eight years. The company’s facial authentication as a service is also in use at financial institutions, hospitals, biotech laboratories, sports arenas, and other sites.
Some of its customers include Scott Data, LAFC, Daytona Airport, and Air Products.
“Shrinkage” has been a major problem for warehouses, with Home Depot reporting $100,000 in theft from a single organized crime ring last year.
“One-third of the Fortune 100 companies are our customers,” said Gaydarzhiev. “We’re looking ahead to a bright future, as enterprises and governments need tighter security. Physical security is catching up to cyber, with a revamp of legacy infrastructure.”
How is Alcatraz AI applying the latest advances in AI to physical security?
“Biometrics traditionally takes a sample from an image, turns it into a bitstream, and compares it later with a scan,” said Frederick. “The problem is that reference template stays the same, but people change — they could lose weight or grow a beard.”
“With the advent of machine learning, we can put processing on the edge and update the models over time so the device works as well on Day 1,000 as on Day 1.” he said. “The Rock also use 2D and 3D data and best-of-breed NVIDIA processors. Our false-positive rate is 1 in 100,000, validated by NIST.”
“We’re also doing cool things around generative AI and contextual control,” Gaydarzhiev said. “It’s like self-driving levels of autonomy — facilities are trying to make decisions based on camera data that affect the physical world. We’re training for Level 4 or 5 for ‘tailgating.'”
