Sensing a Disruption in the Home – Security Sales & Integration

Artificial intelligence (AI) and video analytics, although extremely popular and important, is only one piece of the “occupancy” puzzle. Understanding the layers of AI and various methods of building a comprehensive proposition is similar to the fundamentals of building a comprehensive intrusion system.

The fundamentals still remain in that the most basic intrusion system would include a control, monitoring and some motion detectors. The next step for added security may be glass-break detection and then some window and door contacts for the perimeter. Going further, one may add some internal and then external cameras. Going even further, the cameras and intrusion system would integrate toward a remote video and verification structure. You get the point — it’s like building blocks to design the best and most comprehensive system for the situation at hand. It is not a one-size-fits-all scenario.

The type of AI that is probably the most popular, useful and meaningful in the security industry today is incorporated into video analytics on a video monitoring system. Although AI video delivers great value, it also has some real-world challenges. Challenges that must be considered — and are typically addressable by further investment — include storage capacity, network impact and processing power required to get insights. Additionally, the real-world performance of facial recognition is still suspect and raises a variety of privacy concerns.

There has been a lot of hype (and rightfully so) the past few years about AI because of the power of the new capabilities enabled through its use. Because of all this buzz, it’s often hard to differentiate what’s real from what’s not. Today, nearly every company claims to have some sort of AI.

From a high level, there are two main types of AI: “narrow AI” (an AI that is trained to solve one specific problem as accurately as possible) and “general AI” (an AI that can solve any generic problem, and can think critically and reason about any topic like a human would). Almost all mainstream applications of AI today are narrow AI, no matter how much they may look like general AI.

As of this writing, humanity has yet to create a true general AI. However, we have been able to solve a lot of real-world problems with narrow AI, including many in the security industry. Here are a few that are particularly compelling:

• Facial recognition in video
• Object recognition in video
• Optical character recognition (OCR) in photos and video
• Speaker/voice recognition in audio
• Anomaly detection

Now, AI may be about to allow the industry to unlock additional capabilities, which could have profound impact on the industry: radio frequency device detection, and high-fidelity occupancy data.

New Frontier of Specific Occupancy Data

Truly understanding “who’s in the home” at a high fidelity and high accuracy is a game-changer for security, smart home applications and many other industries. One example of this technology is the HomeAware platform developed by Ubiety Technologies.

Some innovative technologies are emerging that help understand whether or not there is a human present in a monitored area, even if they don’t identify who it is. Consider the following.

WiFi sensing/WiFi motion: Uses existing WiFi waves in the home and watch the WiFi waves as they “bounce” off of objects in the home. More advanced than traditional PIR sensors because it can determine human motion vs nonhuman motion (i.e. no more false positives when the wind blows, shadows change or when a fan moves). An example of this technology is Origin Wireless.

Depth-sensing/“point Cloud” approaches: Use of permanent hardware mounted near the ceilings of monitored areas (and sometimes over entryways/exits) to look top-down and create a 3D point cloud through IR, radar, lidar or other technologies. An example of this technology is Density.

Both of these technologies give good accuracy of recognizing people versus pets or other moving objects, and in many cases offer high enough accuracy to determine pose (sitting vs. standing vs. laying down) and human actions like falling, breathing, etc. These solutions are still anonymized — the system doesn’t yet know exactly who the person is.

Presence detection is an approach to leverage one of many technologies and collaborative integrations toward security and life-safety detection and monitoring. Ubiety’s HomeAware enables homeowners to quickly monitor for the presence of people via wireless device detection without reading, watching, or collecting private information. Image: Ubiety.

In addition, RF device detection is a new generation of emerging capabilities, based on high-fidelity accuracy and understanding of exactly who is present in a monitored area. One new and interesting platform is the aforementioned Ubiety Technologies. It passively listens for signals from RF-emitting devices that people carry, and gives context and insight about what is happening in that area based on those devices.

Ubiety has also built an AI-based, categorization and identification platform that takes the raw RF noise and gives actionable insights about what is happening in a monitored space, opening up a number of new use cases that the industry has not been able to realize in the past. This can be leveraged to create ambient computing “smart home” and automation applications — no need to ask Alexa to play your favorite song anymore, lock the doors when you leave, or arm/disarm the security system; the house can tell who exactly is present versus away, and can take those actions autonomously.

This would mark a significant shift in home automation from the “digital butler” to true home automation.

A Rapid Trajectory

Presence detection has been well-covered in industry publications and industry events. The term does not refer to a specific technology, rather it is an approach to leverage one of many technologies and collaborative integrations toward security and life-safety detection and monitoring. Because it’s easy to misunderstand the ecosystem and variables, outlining the most important takeaways is key.

Additionally, it is essential to understand the value, and especially how it relates when these technologies collaborate with other new and even legacy systems and platform technologies.

As a result of the collaboration between Ubiety and Rapid Response Monitoring Services, when a burglar alarm is received a central station can see specific occupancy data of which residents are at home. Image: Ubiety Technologies.

As previously noted, presence detection and the AI associated with the various platforms has progressed significantly the past few years. One subfield of AI is machine learning — essentially the capability of a machine to imitate intelligent human behavior. AI or machine learning systems are often leveraged by video and presence-detection platforms to analyze datapoints and activity and perform complex tasks in a way that is similar to how humans solve problems.

The interest and precision in presence detection continues to grow. A segment of the players has leveraged and improved existing technology, integrating this with brand new, state-of-the-art AI to deliver value propositions that are arguably disruptive to the security, safety and smart home channels. Other presence detection players include Aqusense, Cognitive Systems and Intellithings.

It is not so much the sensing and cataloging the parties that are present within or near a premise that brings immeasurable value — it’s the AI that analyzes the data and translates and delivers usable information. Understanding who is or is not within the proximity of a premise at any time during an alarm event, or during a certain period of a day by a monitoring center, first responder or a homeowner is priceless.

Just as these platforms leverage different technologies and metrics in delivering these services, understanding that AI included in presence-detection platforms as well as video AI, all bring great value in a host of different scenarios and depending on specific goals of those scenarios. Although some of these platforms are designed to perform as a standalone, in many cases these platforms are designed to perform in collaboration with other systems such as traditional intrusion alarm systems, medical alert systems, video systems, etc., when a more comprehensive system is desired. The future will almost certainly require that all these systems learn to work together.

Leveraging AI for the delivery of a more comprehensive monitoring service is on a rapid and deliberate trajectory. While we see huge impacts now, this is only the beginning. The effect that this will have on how security and safety is viewed will be forever changed. History has proven that information is golden. This is especially true when it pertains to alarm monitoring and detection.

The Next Level

What follows are just a few examples of how monitoring and detection are being taken to the next level.

Home insurance telematics: Telematics data (collected from OBD2 ports or through mobile phone apps) revolutionized the auto insurance industry. High-accuracy home occupancy data about which residents and visitors are in a home is a new “telematics” dataset for the home. The next revolution may be more accurately pricing home insurance policies based on presence detection, or even the RF-emitting devices in the home. This pertains to residences, businesses, and rental and vacation properties.

Accurate prosecution of crimes based on objective evidence: The ability to detect usable information is extremely useful in the prosecution of a crime. Information such as cellular subscriber IDs or device IDs are the most obvious. Many of these identifiers rotate over time and don’t contain personally identifiable information directly but can be used by law enforcement to identify suspects in more serious crimes. This is highly useful for residential, commercial and public safety applications like in cities or schools.

RF device detection and AI working in tandem: Dynamic response flows from central stations when a burglar alarm is received. The central station can now see the specific occupancy data of which residents are home. The automation can dynamically match this against the call list on file and if there are six people on the call list but the first four aren’t home, the call list is delivered to the operator with a new prioritization that goes straight to the fifth person who is present on site and can verify or cancel the alarm quickly. This use case has been pioneered successfully with the collaborative effort of Ubiety and Rapid Response Monitoring Services.

If you enjoyed this article and want to receive more valuable industry content like this, click here to sign up for our FREE digital newsletters!