The solution to this problem came to me during my work on the game called Soldiers: Heroes of World War II. I developed a movement system for tanks and other vehicles for this game.
As soon as we started providing our video recognition system to customers, it became necessary to monitor its stability. Without seeing the system parameters, it can be difficult to understand what is happening, and we had to make assumptions about the nature of the failure guided only by external signs. This was inconvenient and didn’t allow us to understand the cause of the problem and be able to respond to it fast enough. Therefore, we decided to install a monitoring system.
A good idea is not enough for a startup to succeed. It is necessary to have a good understanding of your goals for the business as well as the methods by which they can be in order to get onto the market.
In the beginning of April 2019, we completely overhauled the InvariMatch interface, making it more convenient.
At the moment, self-driving cars can help drivers, but they only reach the third level of autonomy at best. That is why the Tesla manual states that a person has to keep their hands on the wheel while driving, so that they can take control of the car if necessary. The autonomous driving technology is not reliable enough as of now, and a failure in one of the cars subsystem may lead to an accident. But, even at this stage, the autopilot is able to predict and handle a number of dangerous situations much better than a person would.
Modern self-driving cars are equipped with a variety of sensors that allow them to “see” and navigate on the road with ease. Let’s take a closer look at them.
People have been dreaming about autonomous vehicles long before the car was even invented. Only a few decades after Henry Ford had started the mass production of low-cost cars, people wanted more once again.
As a result of optimizing and adding several new features, InvariMatch could now work in a cluster of machines and was much faster at processing search requests than before. It worked fine for the most part, but we still had to deal with occasional system failures.
As the video database of InvariVision grew, searching for similar videos in our system required more RAM and processing time. To solve this, we could’ve added more RAM. However, instead of doing that, we chose the cheaper and more practical alternative of unloading a portion of the RAM onto an SSD.
The InvariMatch system’s original architecture, where all video processing took place on one machine, worked well from the beginning. However, we understood that the system would fail under an increased load and thought about optimizing it. In 2016, one of our customers needed to install InvariMatch on a cluster of several machines. The system worked well at first, but after some time we faced several unexpected problems. We had to improve InvariMatch urgently, and these are the steps we took to achieve that.