In today's article we will talk about the most «mysterious» and futuristic management of CJSC «String Technologies» – about Intelligent Systems Administration.
You could ask what an intelligent system is. – «Brains» would be the simplest and most accurate answer.
The engineering company follows global trends, so the developers have not spared the control over the transport complex. The ability of motion control with the help of artificial intelligence was originally laid in the string transport.
In the intelligent system, which is being developed by the specialists of the administration, there is the whole complex of solutions, which not only remotely control the motion of modules, but also independently make decisions what to do in specific situations on the set algorithms. Recognition of objects on the track using machine vision, various sensors on the track structure and transport module, an interactive system of interaction with users – this is an incomplete list of control developments.
We asked Yevgeniy Rodchenkov, the Head of Intelligent Systems ASdministration, to clarify us about this complex and interesting topic.
Good afternoon, Yevgeniy! I would like to start with simple questions, although I foresee that the answers will be more than the original and unusual ones, as we will focus on such high-tech things as artificial intelligence, neural networks, blockchain and so on. So, could you tell us what an intelligent system in our project is?
The Intelligent system is a technical and software environment, or as it is now commonly called, a cyber-physical system, capable of solving problems at the level of mathematical, clear logic, using the analytical capabilities of neural networks.
The structure of the intelligent system consists of three main blocks: a knowledge base, a decision-making mechanism, and an intelligent interface.
In fact, intelligent systems are a group of sciences that are united under the name of «artificial intelligence» that contains the algorithm capable to autonomously make a decision.
Okay, and what exactly does your Administration do?
We develop and implement control systems using artificial intelligence. The main directions we are engaged in are as follows:
- development of machine (technical, computer) vision system, the task of which is providing information safety of the transport complex (TC);
- development of TC control systems, for automatic (including autonomous) or automated control of transport modules, elements of stations, track structures, switches and other systems;
- development of blockchain platform for secure TC control;
- development of web platforms (sites, applications) for TC control.
Is there any division on departments? What kind of specialist work in your Administration?
There are only 5 departments in the Intelligent Systems Administration: analytics, technical vision, development, web development ones, and there is a new department of advanced development and hardware solutions, which has been recently formed. We have a staff of 28 people. We also cooperate with several companies in the field of control systems analytics and cryptography.
Our team consists mainly of programmers, but there are engineers, scientists, system architects, electronics engineers, project managers, designers, testers, analysts – a complete set for the successful creation of intelligent systems! And due to the rapid growth of the company and the progressive expansion of the line of new products, we are constantly adding new specialists to the team.
Interest in automatic control systems is growing from year to year, and different corporations «try themselves» in this area, as digitalization and automation are trends, which are being tried to follow. The Skyway transport developers also understand the benefits of these processes, therefore, equip their vehicles with more and more «smart» technologies.
So, «eyes and ears» are what the SkyWay transport means already have, thanks to high-tech machine vision. Special sensors read information that occurs around, and the intelligent system based on the analysis of the neural network, detects objects, recognizes them, identifies and makes a decision, depending on the type and situation. The control system «draws conclusions» and transmits the command to action to the on-board system. The specified motion trajectory is recalculated, and a new route task is created. The data enters the database, where automatic schedule changes are instantly made and are stored in the form of accumulated experience.
Could you tell us more about the departments, who is engaged with what?
Every department has its own task list.
The Analytics Department collects and forms technical requirements, together with system architects develops the architecture of the system, justifies the composition and structure of subsystems, components and elements, forms technical specifications.
The staff of the department provide support for the product throughout its development cycle and monitor the compliance of the parameters of the developed product with the customer's requirements.
This department also participates in the development of concepts for the construction of intelligent control systems, monitors the unification and standardization of software and hardware solutions, analyzes the world's leading trends in the development of intelligent systems in the transport sector to implement the best practices in our technology. The department is also engaged in optimization of business processes of management.
The Technical Vision Department deals with the detection, recognition and classification of objects on the basis of information from optical and radar sensors, including the use of sensor fusion technology (combining information from various sensors to improve measurement accuracy).
The specialists of this department are also responsible for the development of the subsystem for monitoring and identification of emergency situations.
The Development Department deals with the full cycle of software development, from idea to implementation. The employees of the department have developed mathematical algorithms for assessing the situation during the vehicle's motion, its positioning on the track, automatic generation of route tasks on the basis of the current, real situation.
The specialists of the department created an emulator of the vehicle's movement, which can be used to conduct virtual testing without the participation of the real vehicle, the simulation of different algorithms and events during the movement.
As one of the actual tasks of the department, we consider the use of big data technology (big data), as for the effective operation of intelligent control systems, there is a need of multidimensional statistics.
The department also develops infrastructure software for dispatching, online services, for passengers, TC users. All the accumulated knowledge and developments are added to the knowledge base, and these developments can be used in the future and in other projects. We do hope that they will also find application in the SpaceWay space program.
The Web Development Department is a separate unit and is responsible for the development and maintenance of sites.
The Advanced Development and Hardware Solutions Department is a newly created department that develops new and prototypes existing technologies in terms of hardware and software (controllers, onboard systems computers, sensors).
This department was formed from people who were previously engaged in the search and development of new approaches and solutions. If we do not find something suitable for us in the market, then we invent, think over and introduce ourselves.
In this way, there was developed, for example, the module standby mode which monitors a «sleeping» vehicle, controlling the battery charge level or other important parameters of the module. If necessary, this unit performs remote transfer of the vehicle from the standby TC («sleeping») to the operating mode, and back, and gives commands to perform the specified actions.
The department is also engaged in research and implementation of advanced solutions, helps in testing, commissioning, layout, software installation. Together with the designers, the type and technical characteristics are being worked out, which then are passed over to the TC layout.
You can conclude that each department works both within its own tasks and within the office as a whole, often assisting each other in solution of related tasks. I believe that in the current structure the Administration functions as an effective, unified organism.
Every second, on the basis of global data, the neural network generates a portrait of a person who does not exist.
However, it is worth remembering that the neural network cannot be used as a decision-making system, it can only be used for analytics when making a decision. There is a certain danger in using a neural network: if it is wrongly trained, it can cause a lot of damage.
For example, in one experiment, it was used to take part in and win the game faster than a person can do it. And neural network has successfully done it. But, to tell the truth, it has won with not quite «fair» methods: using bugs in the game, not the speed of clicking on keys or other «human» methods. That happened because it had been programmed for results, without taking into account the fact that the bug is dishonest behavior, violation of the rules, and it is unacceptable for solving problems.
But it is the neural networks and big data analysis that are now able to give more accurate and faster answers than a human or other mathematical system would be able to find.
Could you tell us about the global development prospects?
We have chosen several directions that we are actively developing and implementing today. Here are some of them: artificial intelligence, automation of the transport complex, big data, provision of digital services, including monetization, and integration of external services with our systems.
The efficient and safe automatic control of the transport complex (TC) is the main task facing intelligent control systems. Making decisions on the optimal transportation of passengers and cargo will be the task of artificial intelligence. Such a system will choose independently how best, faster and easier to transport people or goods.
As for the safety of TC management, here the program will «orient itself on the spot»: what to do in different situations, calculate the solutions and choose the best, which will allow to avoid dangerous situations.
With regard to digital services, it is all about providing services to users of SkyWay information systems. And by integration, we mean full integration of our software with external digital services, merger or symbiosis, depending on a particular situation.
What exactly is currently under development?
That's a certainly interesting question! But I'm afraid, I should write a book to tell you everything about it [laughs]! Our Administration is involved in many processes of the company. We are engaged in intelligent control systems of all transport systems (modules, tracks) and participate directly in the development of potential international targeted projects. The list of tasks is incomplete, of course, as many things are not so easy to tell about.
The most important thing about transport is its safety. Unmanned vehicles have pretty simply divided people into those who «for» them and «against» them. But do not forget that the first tests of any new technology are accompanied by failures, and this is a normal process of development.
The developers of the string transport set initially a goal of decreasing risk of accidents (i.e. road accidents) up to minimum, and that is why the transport system of the «second level» was created: removing the active transportation of people from the ground onto a separately dedicated «line», the string rail, thereby reducing the risk of collisions of passengers, cyclists and other traffic participants.
By equipping vehicles with an automated control system, we remove another factor that leads to tragic errors: human one.
So, it turns out that we train the vehicle to think and make decisions intelligently, but we do not include in the algorithm those human «errors» of thinking that can lead to a mistake, leaving only a clear and unconditional logic.
What has been implemented in practice yet?
I am proud to say that much has already been done.
There has been developed and is being implemented a system of high-precision positioning and acquisition of vehicles' movement parameters on the track structure and stations, based on the integration of information (Sensor Fusion) from RFID readers, Hall sensors, GPS/GLONASS. It will also be supplemented with information from optical and radar sensors, including the use of digital maps.
The object recognition system, adapted to the peculiarities of TC functioning, and the training system are being Implemented. We have implemented an algorithm for identifying potentially dangerous objects against the background of TC infrastructure elements, which is highly relevant for radar sensors. We have also developed the functionality for remote monitoring of the vehicle's condition in working and «sleeping» mode. We created a player for neural network training, thanks to which more than a dozen types of objects are currently recognized with a probability close to one. Work continues on improving the accuracy of recognition and other objects, the total number of which exceeds 90. Our experts have developed and tested, such a seemingly «trifle» as a thermal skin for heat-sensitive equipment, thereby, adapting the vehicle to different climatic conditions.
Thanks to the machine vision system, the vehicle, when detecting potentially dangerous objects, changes the parameters of its movement, up to a complete stop and adjusts its route task in real time.
We have created a simple and intuitive interface which allows to build a route task, taking into account the restrictions on the special sections of the rout; to specify stops for boarding/disembarking passengers, loading/unloading cargo.
The created systems were integrated into the vehicle, and we are training it to make decisions and act in a real situation. Now we have an even more accurate positioning system based on Hall sensors, RFID tags, GPS positioning. The developed systems work in conjunction with the on-board control system, to which we transmit the motion parameters and obtain information about the TC condition. In case of failure of our systems, the on-board control system based on the information previously transmitted to it, should ensure the transportation of passengers/cargo to the destination.
Thanks to the created vehicle emulator in a three-dimensional model, we can create and test route tasks for different conditions: length of routes, passenger traffic, different number and types of modules.
We can say that today we are at the stage of implementation, improvement and training of our intelligent systems: positioning systems, machine vision systems, control of vehicle and transport complex systems.