The surveillance industry is on the brink of a quiet revolution, writes Bilgay Akhan.
In recent years, to name a few applications; drones have become our toys, Amazon is trialling its first drone delivery in the UK. Farming could be revolutionised with drones spraying pesticides. How long will it be before NHS launches its drone army that will take defibrillators to heart attack victims? While all these are changes are taking place in very different industries, are there plans to deploy drones in surveillance? For now, few and far in between; but this is about to change.
Let us just explore briefly a patrolling application in a hotel with large grounds. In a classic surveillance scenario, CCTV cameras are placed in strategic locations, some locations using PTZs. Most borders have physical fences and some critical areas have electronic intrusion sensors. Overall the site appears to be well protected. As additional security, guards patrol the site on foot.
Now let us imagine a scenario where drones are deployed for the same site. The hotel has a swarm of drones being charged in a base station building near reception. Each drone is programmed to fly automatically on a pre-planned route. Drone one: Derek is tasked with patrolling the south east fenced borders, Drone two: Emily is tasked with patrolling the open air swimming pool and surrounding areas. Drone three: Jack is tasked with patrolling a part of the golf complex. Other drones work collectively with Derek, Emily and Jack to patrol other areas of the complex but we will concentrate on these three.
Derek is loaded with a thermal and a day camera as well as a proximity sensor. Derek also has a fail-safe mechanism; if the battery charge falls below 30 per cent, it returns to base automatically. Derek takes off around 11pm as planned, to patrol the fenced borders. While flying over the top of the fence, it processes images it captures. If it senses an intrusion by humans, these images are immediately relayed back to the monitoring centre. The monitoring centre has the ability to control the drone manually. As soon as the centre receives a warning message and images from Derek, they spot a man trying to climb over the fence. The centre takes manual control of the drone and using Derek’s loudspeaker warns the intruder, taking close-up HD pictures of the intrusion. On this occasion, the intruder runs away. The centre now relinquishes control back to Derek. Derek senses it is running out of battery, and the drone automatically returns to its base building, flying in through the specially designed drone hatch.
Emily starts a patrol at 10am around the open-air swimming pool. As it is hovering over the pool at a safe height, it notices a little girl paddling vigorously. Emily’s on board image processing algorithm is programmed to detect several drowning scenarios. This particular scene looks like a case of drowning to Emily. She sends a warning a signal back to monitoring centre, with the image. The monitoring centre takes control of Emily and using the on-board PTZ camera has a closer look. It does look like the little girl is playing a game with her dad in the pool. Centre relinquishes control back to Emily and lets her get on with her patrol, while sending a guard round to the pool just to make sure.
Jack takes off at 3pm to patrol a part of the golf area. Ten minutes into the flight, using its on-board image analysis, it notices a “man-on-ground” and sends images back to control centre with a warning message. The monitoring centre plays back and examines the images and launches the first-aid drone to the scene. The first aid drone lands and drops the first aid kit, flying back to base. In fact one of the golfers fainted. One of the group takes charge and using the first-aid kit, helps him come round while the monitoring centre is watching from the control centre. Nothing serious and thanks to timely help by the first-aid drone, the incident has not developed into anything more.
While the above paragraphs may sound like a scenes out of a science fiction movie, this technology is here now!
Technology
A typical surveillance drone needs thermal, daylight cameras, 3D-GPS, 4G and wifi connectivity; which exist today. Most drones can capture and relay images but on-board image processing is not widely available on drones yet.
Cloud connectivity is one of the essentials for surveillance. Drones could be attacked and may lose their on-board capabilities but if they are cloud-connected, surveillance information will be safe. While cloud connectivity exists in some commercially available drones, capabilities and bandwidth can be limited. On the other hand, companies such as airware (https://www.airware.com) are developing chipsets and cloud connected operating systems specifically for drones. AIRWARE has raised $30m investment with Cisco as one of its early investors. AIRWARE only collaborates with high potential partners while it is developing its Application Programmers Interface (API). Such developments clearly demonstrate that drones are expected to embrace many high volume applications.
In a hotel or golf complex type of application, the technology is easier for cloud connectivity. The complex may station many wifi routers at strategic places for wifi coverage. But some applications such as oil and gas fields may require 4.5G connectivity.
Advantages over fixed cameras
Drones have some clear advantages over traditional security cameras and intrusion detection including:
●Full aerial imaging with no blind spots
○360 degree view, no missing details
○Ability to zoom on a scene almost with no limits
●Mobility
○Harder to be deactivated by criminals
○Tracking mode can be used to offer extra security for guard
○Companion mode at night to feel more secure
●Cost
○A drone surveillance system may be more economical than cameras covering the whole grounds
While drones, on their own can not provide a total surveillance solution, not only can they strengthen existing solutions and reduce costs but also they have the potential of bringing new dimensions in surveillance. A good example of a new dimension is a drone-based “neighbourhood watch” scheme.
Privacy concerns
Rapid development of drones has developed privacy concerns and these also apply particularly to surveillance. Imagine police deploying drones as “Neighbourhood-Drone-Watch” in your neighbourhood. I am sure while some will feel more secure, some will certainly have privacy concerns. Drones with thermal imaging cameras and advanced image processing can be used to track individuals easily. There may be many ways to overcome these privacy concerns but more debate is needed by wider society to take it to acceptable conclusions.
Challenges
There are technical challenges for drones in surveillance. Other bits and pieces are needed to complete the puzzle. To name a few:
– On-board image processing
– ‘Sense and Avoid’ technology
– Intelligent tracking
– Flight under all weather conditions
– Propeller noise control.
Another important issue that deserves a mention is “energy”. Most drones have a flight time limited to 30 to 60 minutes. When they are loaded with cameras, this comes down to 15 minutes or so. It is reasonable to expect that with recent advances in battery technology, this problem will probably be solved earlier than others.
About the authors
droneResearch is a technology company specialising in development of application specific drones with a particular focus on the surveillance, healthcare and telecommunication sectors. Visit www.drone-research.com.
Dr Bilgay Akhan is founder and group product manager at droneResearch and is also a member of Texas Instrument Inc. Expert Advisory panel.
[email protected]
Kerem Kayabay is a senior product manager at droneResearch and is also a Ph.D. candidate at METU; Middle East Technical University, in Ankara; www.metu.edu.tr. His research interests include drones and cloud applications. Email [email protected].
