Blog
GREAT EXPECTATIONS: ACCESS CONTROL AND AI
What can we expect from AI? This question is currently is on the minds of almost everyone, including the access control community. The broad consensus is that we should expect the unexpected. We agree completely, access control is not an exception. However, there are a couple of features already taking shape on the horizon. So, what are the areas where the access control industry can expect to benefit from AI in the near future?
UNUSUAL BEHAVIOUR DETECTION
One area where AI traditionally excels is pattern recognition. We expect the new generation of machine learning to be able to continuously analyze the live stream of access control events and be able to single out the “unusual” cases in real time. For example, someone appearing at unusual time or at unusual place. Or a bunch of events appearing with unusual frequency or clustered in an unusual way. The system could then learn and improve further from human feedback. For example, the system could learn that certain unusual patterns are of no concern.
Currently there is no straightforward way to extract such information from the access logs, especially in real time. AI detection could prove as excellent additional security feature, particularly for larger access control systems handling hundreds of doors and thousands of daily users.
SYSTEM SETUP AND MANAGEMENT
Access control systems, especially larger ones are often quite complex to set-up. Managing groups of users with identical access rights, users with special access rights, door groups, special doors, door schedules, user schedules, group schedules, and other parameters can be overwhelming, especially during the initial deployment phase when the most work is done while the administrators are least experienced.
This is where AI can provide a natural language interface, a chat-based setup and reporting tool. For example, the user may enter something like this:
Allow access to Room 122 to all users from Unit XY, except from 10PM to 6AM and on Sundays.
List all users with access to any room from the area Floor 23 on Sunday midnight.
A well-designed chat-based interface would simplify the setup process, eliminating the need for deep menu diving and parameter searching. It would enable administrators to make changes smoothly and efficiently, such as adding new users, assigning access rights, and defining access schedules.
TANGENTIAL: TAILGATING DETECTION
Although video surveillance falls outside the realm of access control, AI-driven video analytics have the potential to address the challenge of tailgating. Tailgating refers to when multiple individuals enter using a single person’s access badge. While camera-based tailgating solutions already exist, we believe that the real breakthrough is still ahead of us, but not too far in the future. All we need is a powerful camera platform capable of running advanced AI algorithms in real-time. By developing a reliable, self-setup, self-learning tailgating detector, we can enhance security and prevent unauthorized access.
CONCLUSION:
By harnessing the power of AI, we can expect improved detection of unusual behavior, simplified system setup and management through natural language interfaces, and enhanced security measures against tailgating. As AI continues to advance, access control systems will become more efficient, user-friendly, and effective.
