Leo Ryzhenko, the founder and CEO of SMP Robotics on how the company got started and where it’s heading

When surveillance cameras first came on the market, they revolutionized the security industry. One can pass a security guard unnoticed, but the surveillance camera’s “always on” ability makes it nearly impossible to go undetected. But as surveillance cameras were more widely embraced, it didn’t take long for users to realize that monitoring the video feeds was no easy task. It’s not uncommon for security guards to review the footage from 20 to 50 cameras and as the number of cameras an operator is tasked with monitoring increases; their ability to do so effectively decreases.

Eventually, this problem was resolved by video content analysis (VCA), now more commonly referred to as video analytics. Using a combination of algorithms, video analytics analyzes captured video in real time and presents alerts about whatever the application is programmed to identify. In early versions, this was primarily motion. But as you can imagine, things move all of the time, especially outdoors. As a result, the first uses of video analytics resulted in a high number of false alarms. Stronger wind and rain would set off the alarm. Eventually, the large number of false alarms caused security guards to ignore alerts, which defeated the purpose of having video analytics in the first place.

As most of us in the security industry know, outdoor video surveillance systems can be quite expensive. For instance, a mile of metal fence equipped with cameras can cost up to one million dollars and traditional fixed surveillance cameras have an inherent flaw: blind spots. When a fixed camera is pointed right and the incident occurs on the left its not gong to be captured. There is lack of dynamics and can be difficult to merge the footage from different fixed cameras. That’s how we came with an idea of a moving camera, and from that to a mobile security robot. The moving robots are easy to implement, they provide a comprehensive overview of the area, but the intruder recognition problem still remained.

Once we got to that point, we realized that it was necessary to teach the robots to recognize the violators or the situations when one should send a response team. This is the most difficult task that company faced. We worked on this problem for more than two years. Once NVIDIA deep learning system and mini supercomputer Jetson TX2 became available that dramatically accelerated our development. Now S5 security robots are able to recognize people, faces, cars, license plates, uniforms, gender and even detect an unusual behavior. However, a robot needs to be trained but there are no systems or methods for that. The robot enhanced with artificial intelligence does not have a “button” that you can press and it just starts working. Therefor AI operational training systems are one of the main tasks our company is currently working on.

I have an IT background and 20+ years of experience in data mining. I began to study the security topic 10 years ago; he was particularly interested in getting the data processed quickly and effectively. In 2008 – 2009 my team and realized that combining robots and AI was the right way to go about this.

The technological progress in the field of autonomous robotic movement over rough terrain, combined with the processing of information from the surveillance cameras, and later the introduction of AI, made it possible to create the autonomous security robot. With its excellent team of engineers and experts in different areas, SMP Robotics solved all these problems in the S5 Security Robot.

Partnership with NVIDIA and implementation of the Jetson TX2 made a breakthrough in the way of improving the guarding and patrolling methods. The robot does not replace the human guard, but thanks to the AI and object and behavior recognition, S5 robot becomes a reliable help in patrolling various areas. The system allows processing and analyzing videos from an unlimited number of cameras without human involvement and, depending on the result, it makes a decision to notify security.

The strong surface grip allows the robots to surmount level differences, bumps and ditches. The robots can function in various terrain conditions: road, off-road, grass, gravel and snow. The robot can operate in all types of weather conditions (rain, snow, fog, smoke, etc.), with temperatures ranging from -4F to +113F. The robot is able to collect and process information in areas of natural disasters, explosions, collisions to detect and identify objects, before the human forces are sent to the scene of the accident. HD cameras with 360 degrees observation allows capturing videos of the terrain and transferring it in real time to a Security Operations Center (SOC).

The first robots enhanced with AI from NVIDIA Jetson TX2 were shown at the ISC West conference that took place in April in Las Vegas and then at the NVIDIA GCT Conference in May 2017 in San Jose. The new intelligent robots will become available for delivery to end-users in October 2017. Along with SMP’s master distributor Robotics Assistance Devices, we are engaged in creating a robust service and maintenance system in the largest metropolitan areas of the US and Canada.

We partner with the leaders of Artificial Intelligence Industry and constantly interact with security companies to identify their needs. It gives us the opportunity to determine the product development roadmap. SMP Robotics is actively expanding our production facilities and soon will be ready to supply significant number of S5 robots to our end users on a monthly basis. We are also actively expanding our global distribution network and always one step ahead and are ready to solve the most difficult tasks in the field of security and patrolling!

Leo Ryzhenko, Founder and CEO of SMP Robotics.