There has been a lot of talk recently about Artificial Intelligence and what it can and cannot promise. One thing is for sure, the shoots of intelligent software are everywhere and none more so than in Merlin.
Our software is designed to integrate all of the services you find in a modern building, town or city, even a country. It scales really well. It encompasses all of the systems that many security companies rely on for their bread and butter, for example the monitoring, control and despatch of Police and other responders when intruder alarms trigger: technology that is somewhat dated but still supported by the mainstream industry – to survive you have to have this capability. For over eight years now Merlin has additionally given security companies the opportunity to give their clients the best chance of protecting their assets by tying together all of the CCTV systems they have deployed, irrespective of the type and model used. For the control room analysts the interface is common, one display type for all types.
As well as this commonality Merlin sees the incoming alarm signal and pulls recorded CCTV from the device, whether its an edge device like a Mobotix camera, VSaaS, a DVR/NVR or even a VMS. This unique capability is vital in the modern control room as it is unlikely that the analysts can immediately go live to a CCTV system when an alarm occurs; it takes time for the system to connect and set-up the streams and there is always the chance that the analyst has a queue of events to deal with. Even a delay of twenty seconds gives the control room little chance of seeing what caused the event in the first place.
Now, in addition to this smart handling of CCTV, bearing in mind that Merlin may have more devices connected to it than any other control room software, we introduce threat and risk assessment in real time. Merlin will learn, using it's own intelligence and memory, the pattern of alarm triggers and will pattern match. This clever piece of code enables Merlin to think for itself. Giving the analysts an indicator of an event that may be real, picking it out from the noise and clutter of the busy control room. It is truly artificial intelligence, Merlin thinks for itself based on a strictly controlled and managed set of rules built from years of experience.
Used properly MerlinAi will control an active video wall, finalising the concept of 'dark screen' and showing to the analysts cameras and sensor information that it finds more interesting than the normal inactive display of hundreds of pointless and ignored CCTV images.
MerlinAi uses historic data, taking its reference from what happened in the past, how events matured and how the analysts dealt with them. It takes input from weather, non associated activity and short term memory of recently dealt with incidents and probes. For perimeter protection of a large high value location like a prison, security compound or factory, where sensors are notoriously and, by demand, sensitive this intelligence is invaluable. Triggers related to weather and environment plague the analysts, MerlinAi now gives them a fighting chance without the requirement to detune sensors.
In the commercial control room, even looking at something as mundane as an intruder alarm you now have the capability to look at the signals coming from every sensor, each time a sensor triggers that data can be gathered not just in the alarm equipment but in the control room. Using the SIA and CID protocols properly, as designed in the 1990's before they were abused and neutered. No longer is it necessary to install two different types of detector in each zone, the system will show the operator what the alarm is doing, in real time, and boost the risk assessment if it thinks the threat is real. Remembering of course that if CCTV is available the correct cameras, live and pre and post imagery can be made available at the analysts desk – in one cohesive screen.
For control rooms that monitor remote CCTV systems the advantages are even greater. Merlin will de clutter the number of alarm triggers, it will prioritise the events based on risk assessment of the type of trigger it is receiving. As time passes it learns more and gets better, just as the dedicated analyst does right now, she or he has learnt that some sites they protect give more false triggers at certain times than others, she or he has learnt that it may be better to look at something more unusual than that which is shouting too often.
Artificial Intelligence in a limited and proactive way is here and working, it is proven and reliable. It will give the early adopters better chance at giving their customers more reliable detection, faster and more efficiently than non integrated and dumb systems of the past.