Autonomous robot control. Autonomous robot builder. Autonomous Navigation Equipment

Which is used to control it. The autonomous robot has two different control programs. The first program allows the robot to ride avoiding obstacles on its path, to determine their robokar uses two ultrasonic sensors. The second program is the plan of surrounding items using a two-dimensional array. After receiving data from a two-dimensional data array, the robot will know where and what is around it.

Materials:
- Ultrasound Sensors 2 pcs (4 pieces for modernization in the future)
- Servo drives 4 pcs
- Arduino (the author uses the UNO model)
- Bread board
- Wire
- Accumulators 9,6V 2 pcs
- Battery 9V.
- Wheels 4 pcs
- Isol
- Nuts, bolts, etc.

Step one. Mechanical part.
First of all, the robot requires a solid chassis. The article has a photo of a robot, but what a chassis to use and how to do it does not matter. The author did three different options of the robot. The article discusses only two options, as the third was not particularly successful. The first version of the robot had a form resembling a truck. He had big sizeBut had a rather low speed and poorly unfolded. In addition, a large robot is not very convenient to use. The second option was made more thought out, it turned out much less and more compact.
First, the chassis put servo drives, so that there was an opportunity to put on their shafts. The author uses four wheels. If you take powerful servo drives, then you can use two wheels at all. But the chassis at the same time you need to arrange so that there are enough space for batteries, pCB And Arduino.

After installing servo drives, they put wheels. The author installed an additional protection against the wheel in the wheel after the wheel. On the front of the robot, two wheels are additionally delivered, which will be able to help the robot to travel to borders or other small obstacles if he is preserved in them. To reduce friction on the rear wheels, added the isolent.

Next, the battery compartment is installed. The author took the VEX charger, and modified it to power the engines, and not charging the batteries. Now the fee is taken, the Wires Plus and GND are disappeared with it, which will go to the battery charging connector. Then the black wires from two batteries are soldered to the GND charging wire, and the red wires from the batteries to the positive wire charger. Then these wires are connected to the board. After that, the author makes fastenings to install ultrasound sensors on the front of the robot. If you need add additional sensors will need to lengthen the mount.

Step second. Electronic part.
For this step, greater knowledge in electronics will not need. 9,6V batteries are connected in parallel, if the battery compartment is used from the charger, then nothing will be done, as it has already been done. Next, according to the scheme below, all components are connected. It should be noted that depending on the chassis length you need to select the wires, or lengthen them, as they may be lacking up to the board. One signal wire is used for the first and second servo drives, and for the third and fourth different. This is done for synchronous work of the first and second servo drives, because they are located on one side, the same applies to the third and fourth servos.

To add additional sensors or servo drives, everything is done on the same principle - the ARDUINO is connected to the signal wire, GND to the black, and the power of 5V to the red wire. It should be remembered that the GND from the engines must be connected to the GND battery and Arduino.

Step Three. Software part.
For writing code, the author used Processing. For navigation, a two-dimensional array is used, the values \u200b\u200bof 0 or 1. If you enter 1, it will designate the object, it means that the robot will ride only by 0. The code can be downloaded below.

Modern robots, like many decades ago, have a limited list of action algorithms and are useless in difficult conditions without communication with the operator - with strong radiation, under the ground, at sea depths or in space. The sad example of a recent catastrophe on Fukushima showed that no remote control will replace the real autonomous robot.

Most robots are programmed at the level of instincts or are controlled remotely. Autonomous robot without autonomous artificial intelligence is impossible.

The main problem of modern robotics, like many decades ago, is related to the development of decent artificial intelligence. In some cases, you can talk about the successes and even obvious progress. For example, Google's experiments on the creation of autonomous cars without a driver or the Alphago program of the same Google, winning world champion in the game. Or supercomputer intelligence IBM Watson, able to understand the questions and find answers in the knowledge base.

photos

So far, most of the developments in the field of artificial intelligence are not suitable for autonomous robots. Many of them are limited to the selected scope of application, some require non-alone computing capacities. In some cases, both Microsoft's robot, an artificial mind and comes crazy after a brief communication with people at all.

Today, the autonomous robot must understand natural speech and gestures, to think logically, learning and make independent decisions. The perfect autonomous robot, equipped with the necessary sensors, tools and the knowledge base, must listen to the task and without unnecessary questions Start the ravis on its implementation.

Russian "intelligence": universal brains for an autonomous robot

Russian developers have always been famous for a wide view of the problem. Artificial intelligence "Defense", created by the domestic company "MIVAR", was originally developed as the basis for any Types of autonomous robots.

The program logical core of the "intelligence", speaking robototechnical language - this is a logical scheduler, providing robots the opportunity to independently build algorithms and solve problems without people's participation. The difference between the "intelligence" of the robot vacuum cleaner and the artificial intelligence of the autonomous robot excellently explains the slide below, where the difference between the reflexive and logical level is shown.

Three-dimensional diagram of research in the field of artificial intelligence

The work of the artificial intelligence "The Multivoretor" is described by the "Mivo Principle", which means processing multidimensional databases with a context-global model, where data, their logical output and processing process are integrated into a single integer, and all processes occur in real time. Acronym "MIVAR" (multidimensional informational varying adaptive reality, in English Multidimensional Informational Variable Adaptive Reality), consonant with the company's name, with successful development technology, has every chance of entering foreign languages \u200b\u200bin the same rights that Sputnik entered the same rights.

Multidimensional data analysis technology allows autonomous solutions in real time

The main feature of the musnic technology is the extremely high speed of work - up to 5 million rules per second. Thus, for analyzing huge data arrays and the adoption of operational autonomous solutions is sufficiently low computing power. For a minute: a regular laptop can handle a 20-dimensional graph with 150 thousand vertices for 600 thousand ribs in real time! The indicators are so high that, according to Oleg Varlamov, while no one in the world is ready to compete with them.

The "intelligence", also presented in the form of boxed software, can be installed even on the usual laptop.

The "intelligence" is the core of the complex project "Roborum", which can be adapted to any type of autonomous robots. Let's say tomorrow a representative of an exploration company will come and will order a deep-water autonomous shark for Arctic conditions - appearance The robot will change radically, but the brains of the "intelligence" will remain the same, except with additional adaptation to the appropriate knowledge base.

Robotlate Murom-Span: Universal blank for the production of autonomous robots

The integrated system of the autonomous robot for any purpose includes five basic elements. The list includes mechanisms, sensors sensors, computing module, autonomous power elements and artificial intelligence itself.
Oleg Varlamov
president of Miva

The robotics platform "Murom-Span", created by Miv and Intellectual Technologies, is a universal designer: the first four basic elements from the list voiced above are completed in the measure of needs, fantasies and agents. The fifth element, as in the film of the same name, cannot be replaced: this is the intellectual core of the "intelligence".

"Murom-Span" is created as a landfill for the "intelligence" run. Folding anthropomorphic robot with a touch head and self-balancing uniaxial chassis height of 165 cm in unfolded and 80 cm in the folded state will allow to work out the assemblies of the autonomous robot and its interaction as part of more complex complexes.

Specifications of the first prototype Murom-Span: The Ministry of Emergency Situations will be happy

By the way, about reflexive robots vacuum cleaners. In the autonomous complexes, such as "Murom", the role of remotely managed auxiliary mechanisms serving for sensing, clearing and even hiking is prepared in such subsidiaries. Such an assistant can be sent to intelligence, but even the loss of one or several auxiliary robots will not affect the efficiency of the complex.

Combined mechanisms and sensors, controlling the whole brood of auxiliary robots, "Murom" can be part of a more powerful complex. Imagine a powerful autonomous system on the "Kamaz" platform, carrying a dozen "Muroms" of special purpose with hundreds and other auxiliary self-based assistant robots. This is where unlimited space for fantasies of civil and defense customers is revealed!

Externally, "Murom" does not shine a special attractiveness, but the design of the system is fully balanced on autonomy, performance and computing power. Now "Murom" works on five intel processors Core i5. According to Vladimir Denisenko, the director of "Intellectual Technologies", the experiments were carried out with various platforms, including accelerators on graphic maps.

So far five Intel Core I5 turned out to be optimal in the ratio of productivity, autonomy and price, but there is no binding to any particular hardware and software platform. When the need for the platform on the domestic processors "Elbrus", this version will appear immediately.

MUROM-OFF: The carrier on the basis of KamAZ deploys a dozen autonomous robots with hundreds of robot helpers

Fully working version of "Murom" with voice control, Synthesis of speech, manipulators and other functions, developers will appear in September 2016. Today, the "intelligence" can be used by all those who wanted both as an independent product and as an integrated logical component of other control systems - up to the level of the API.

According to Oleg Varlamov, Miva company is open to cooperation with Russian companies, institutions, startups and even single enthusiasts, which the "intelligent" can be provided to the most favorable conditions, Up to free samples.

When it comes to building robots and 3D printing houses, most people present a high-tech mechanism under human control. Such robots at least need one operator and assistants - people who will serve them. Engineers from the Massachusetts Institute of Technology decided to look into the future and developed the prototype of the autonomous robot-builder.

The robot is a self-propelled chassis. In the front of the "Builder" is "hand" - a smart multifunctional manipulator.

For "hand" with a computerized control mechanism, a platform with construction materials is placed.

On the this moment To work out the concept of the robot, it takes a capacity with PPU (polyurethane foam) and foam concrete, from which he builds a dome house using the non-co-formwork technology. First, the outer and inner pamp wall is erected, and the gap is then filled with foam concrete.

On the construction of the domes with a diameter of 15 meters, the robot spends about 10-14 hours.

At the same time, the robot does not need to manage the person and himself, oriented on the land of the marks, decides how to build a construction.

If the solution ends, the robot goes to the database for refueling, after which the construction of the house continues.

According to the developers, the choice of materials and the construction techniques is due to the fact that it is so easier to "teach" the robot to build and make appropriate adjustments in software.

The next step is to use concrete for constructing, which for mobility is modified by special additives.

To work with different construction solutions, a set of spray pumps is provided.

In addition, as the robot improves, they will teach work with reinforcement, welding machine.

And excavator bucket.

According to the plan of engineers, over time, on the basis of the prototype, it will be possible to create a fully autonomous robot.

For example, one of the scenarios can be sent to several builder robots to the remote area, where they interacting with each other, begin the construction of buildings.

Keywords

Autonomous mobile robots / Navigation system / SLAM / fuzzy logic / Linguistic variables / Fuzzy logic output / Multigent robotic systems / AUTONOMOMOUS Mobile Robots / Navigation System / Fuzzy Logic / Linguistic Variable / Fuzzy Logic Inference / Multiagent Robotic Systems

annotation scientific article for electrical engineering, electronic technology, information technologies, author of scientific work - Mikhailov Boris Borisovich, Nazarova Anid Vartanovna, Yushchenko Arkady Semenovich

New methods of management and navigation with robots capable of autonomous behavior in conditions of non-deterministic working environment. The activity of the human operator is reduced to the observation of the functioning of the robotic system and to the formulation of current tasks. In this case, the feedback is added, which may be speech. Thus, the task of management acquires the nature of the person's dialogue and the robot, accompanied by graphic and speech messages. Essential role is playing navigation systemSince the robot must independently evaluate the environment and plan its path, including in the presence of other moving objects in the working area. The automatic solution of these tasks significantly facilitates the task of the operator, but requires the development of an "intellectual" control system of the autonomous robot. Such tasks also include the task of the automatic return of the robot in the loss of communication with the operator, the solution of which increases the reliability of the robotics system. Changing the nature of the operator's activity, which now does not manage directly by the movements of the robot, leads to a change in the nature of the management system, since it should take into account the possibility of perception of the operator and the nature of the solutions taken by it. One of the ways to solve this problem is the use of fuzzy logic both at the perception stage of information and at the stages of planning actions and the adoption of operational solutions. The use of "natural" spatio-temporal relations and linguistic variables Even more approaches the management process to the operator's dialogue with the robot, which allows you to determine the robotic system of this type as a system of cooperative management. In practice, most of the tasks are solved autonomous mobile robotsSuch as a terrain monitoring, radiation and chemical intelligence, fighting fires and natural disasters require the participation of the Mobile Robot Group. Some results concerning group management of robots are also presented in the work.

Autonomous Mobile Robots- Navigation and Control

The New Methods of Control and Navigation of Autonomous Mobile Robots In Undetermined Environmente Are Under Consideration In The Paper. The Human Operator CAN Only Observe The Behavior of the Robotic System and State the New Tasks. The Feedback May Be in Form of Speech. SO The Control Task Acquires a form of a Dialogue Between Human and Robot Accompanied by Graphic and Speech Information. Important Part in the Autonomous Robot Control Is The Navigation System Proving The Robot to Appreciate The Environment and To Plan Its Own Way. Also in Presence of Oter Moving Objects in the Scene. Automatization of Such Operations May Make Sufficiently Easer The Tasks of the Operator By The Intellectualization of Control System of the Mobile Robot. One of these Tasks Is The Robot Return in Case of Communication Loss with the Operator. The New Mode of Operator's Control Lead to New Mode of Control System Which Now Has to Take Into Consideration The Subsibilities of Human to Perceive Information and to Accept the information and to accept. One of the Way to Solve The Problem IS The Application of Fuzzy Logic Both On The Stage Of Parning and Decision. Application of the "Natural" Relations of Space and Of Time Make The Dialogue Similar to the Dialogue Between The Master and Assistant. So Such Robotic Systems May Be Named The Cooperative Ones. Their Practical Applications Usually Demand The Participation of A Group of Autonomous Robots To Fulfil The Stated Task. Some Results In This Direction Also is pretented.

Text of scientific work on the topic "Autonomous Mobile Robots - Navigation and Management"

17. PShikhopov v.Kh., Medvedev M.Yu. Adaptivnoe POZITSIONNOE UPRAVLENIE PODVIZHNYMI OB "EKTAMI, NE LINEARIZUEMI OBRATNOY SVYAZ" YU, MEKHATRONIKA, AVTOMATIZATSIYA, UPRAVLENIE, 2015, VOL. 16, No. 8, pp. 523-530.

18. Martynova L.A., Mashoshin A.I., Pashkevich I. V., Sokolov A.I. Sistema UPRAVLENIYA - Naibolee Slozhnaya Chast "Avtonomnykh Neobitaemykh Podvodnykh Apparatov, Morskaya Radiolektronika, 2015, No. 4 (54), PP. 23-32.

19. Martynova L.A., Mashoshin A.I., Pashkevich I.V., Sokolov A.I. INTEGRIROVANNAYA SISTEMA UPRAVLENIYA AVTONOMNOGO NEOBITAEMOGO PODVODNOGO APPARATA, MATERIALY 8-Y VSEROSSIYSKOY MUL "TIKONFERENTSII PO PROBLEMAM UPRAVLENIYA, DIVNOMORSKOE, 28 SENTYABRYA - 3 OKTYABRYA 2015G, VOL. 3, PP. 191-193.

20. Martynova L.A., Mashoshin A.I., Pashkevich I.V., Sokolov A.I. Algoritmy, Realizueemye Integrirovannoy Sistemoy Upravleniya Anpa, Izvestiya Yufu. Tekhnicheskie Nauki, 2015, No. 1 (162), pp. 50-58.

21. GORODETSKIY V.I., GRUSHINSKIY M.S., KHABALOV A.V. MNOGOAGENTNYE SISTEMY (Obzor), Novosti Iskusstvennogo Intellekta, 1998, No. 2, pp. 64-116.

22. RZHEVSKIY G.A., SKOBELEV P.O. KAK UPRAVLYAT "SLOZHNYMI SISTEMAMI? MUL" TIAGENTNYE TEKHNOLOGII DLYA SOZDANIYA INTELLEKTUAL "NYKH SISTEM UPRAVLENIYA PREDPRIYATIYAMI. SAMARA: OFUT, 2015, 290 p.

23. Innocenti B. A Multi-Agent Architecture for an autonomous robot. Ph.d. Dissertation - Universitat de Girona, 2009, 146 p.

Martynova Lyubov Aleksandrovna - Concern JSC "Electric Appliance"; E-mail: [Email Protected]; 190068, St. Petersburg, Rimsky-Korsakov Ave., 49, square. one; Tel.: +79219411395; NIC "Integrated environmental lighting systems"; D.T.N.; S.N.S.

Mashoshin Andrei Ivanovich - E-mail: [Email Protected]; 197046, St. Petersburg, ul. Small landing, 30; Tel.: +79217632345; NIC "Integrated environmental lighting systems"; D.T.N.; Professor.

Martynova Liubov Alexandrovna - JSC CSRI Elektropribor; E-mail: [Email Protected]; 49-1, PR. Rimskogo-Korsakowa, Saint-Petersburg, 190068, Russia; Nic "ISOO"; Phone: +79219411395; DR. Of ENG. SC.; SENIOR SCIENTIST.

Mashoshin Andrey Ivanovith - E-mail: [Email Protected]; 32, Malaja Posadskaja Street, Saint-Petersburg, 197046, Russia; Phone: +79217632345; Nic "ISOO"; DR. Of ENG. SC.; Professor.

UDC 621.865 (075.8)

B.B. Mikhailov, A.V. Nazarova, A.S. Yushchenko

Autonomous Mobile Robots - Navigation and Management

New methods of management and navigation by robots capable of autonomous behavior in the conditions of a non-deterministic work environment are considered. The activity of the human operator is reduced to the observation of the functioning of the robotic system and to the formulation of current tasks. This is added feedback, which

maybe speech. Thus, the task of management acquires the nature of the person's dialogue and the robot, accompanied by graphic and speech messages. A significant role is played by the navigation system, since the robot must independently evaluate the environment and plan its path, including in the presence of other moving objects in the working area. The automatic solution of these tasks significantly facilitates the task of the operator, but requires the development of an "intellectual" control system of the autonomous robot. Such tasks also include the task of the automatic return of the robot in the loss of communication with the operator, the solution of which increases the reliability of the robotics system. Changing the nature of the operator's activity, which now does not manage directly by the movements of the robot, leads to a change in the nature of the management system, since it should take into account the possibility of perception of the operator and the nature of the solutions taken by it. One of the ways to solve this problem is the use of fuzzy logic both at the perception stage of information and at the stages of planning actions and the adoption of operational solutions. The use of "natural" spatio-temporal relations and linguistic variables even more approaches the management process to the operator's dialogue with a robot, which allows you to determine the robotic system of this type as a system of cooperative management. In practice, most of the tasks solved by autonomous mobile robots, such as terrain monitoring, radiation and chemical intelligence, fighting fires and natural disasters require the participation of the Mobile Robot Group. Some results concerning group management of robots are also presented in the work.

Autonomous mobile robots; navigation system; Slam; fuzzy logic; linguistic variables; fuzzy logic conclusion; Multigent robotic systems.

B.B. Mikhailov, A.V. Nazarova, A.S. Yuschenko Autonomous Mobile Robots- Navigation and Control

The New Methods of Control and Navigation of Autonomous Mobile Robots In Undetermined Environmente Are Under Consideration In The Paper. The Human Operator CAN Only Observe The Behavior of the Robotic System and State the New Tasks. The Feedback May Be in Form of Speech. SO The Control Task Acquires a form of a Dialogue Between Human and Robot Accompanied by Graphic and Speech Information. Important Part in the Autonomous Robot Control Is The Navigation System Proving The Robot to Appreciate The Environment and To Plan Its Own Way. Also in Presence of Oter Moving Objects in the Scene. Automatization of Such Operations May Make Sufficiently Easer The Tasks of the Operator By The Intellectualization of Control System of the Mobile Robot. One of these Tasks Is The Robot Return in Case of Communication Loss with the Operator. The New Mode of Operator "S CONTROL LEAD TO NEW MODE OF CONTROL SYSSTEM WHICH NOW HAS TO TAKE INTO CONSIDERATION THE POSSIBILITIES OF HUMAN TO PERCEIVE INFORMATION AND TO ACCEPT THE NESSARY DECISIONS. ONE OF THE WAY TO SOLVE THE PROBLEM IS THE APPLICATION OF FUZZY LOGIC Both On the Stage of Perception and On the Stage of Planning and Decision. Application of the "Natural" Relations of Space and Of Time Make The Dialogue Similar to the Dialogue Between The Master and Assistant. So such Robotic Systems May Be Named The Cooperative Ones . Their Practical Applications Usually Demand The Participation of A Group of Autonomous Robots to Fulfil The Stated Task. Some Results in this Direction Also is a preceded.

AUTONOMOMOUS MOBILE ROBOTS; Navigation System; Slam; Fuzzy Logic; Linguistic Variable; FUZY LOGIC INFERENCE; Multiagent Robotic Systems.

Introduction Modern mobile robots can independently move in the surrounding space and perform necessary actions Using manipulators. The robot is equipped with a system of technical vision and a complex of information sensors capable of forming an integrated idea of \u200b\u200bthe current situation. The knowledge base of the robot allows him to independently navigate the environment and make decisions on the actions necessary to solve the task. Thus, manipulative mobile

the robot is an "intellectual" technical system capable of autonomous behavior. However, in most tasks performed in advance of not specific conditions and associated with a high "price" error in incorrect actions, the human operator's personnel in the management of the robot continues.

Application Robotics B. different applicationsassociated with solving special tasks requires maximal simplification of the methods of human interaction and robot. The most natural way of such interaction is the speech dialogue. The task of managing the robot on the part of the operator in this case includes a dialogue on a problem-oriented language, close to natural, and monitoring the actions of the robot. The formulation of the management task in this case is modified, since the robot becomes no longer an object of management, but a technical subject-partner, able to independently determine its sighs and a line of behavior in the interests of the overall task supplied by the operator. Role feedback In the dialogue system, speech messages of a robot operator operator who are able to clarify the commands, informing the operator about the current situation, or about achieving the goal.

A special role in solving the tasks of managing autonomous robots belongs to an information and sensory system, which must independently analyze the current situation, plan its actions and at the same time interact with the human operator in a language close to the natural language. It must independently search and detect hazardous items, freely move in a space in which other moving objects can be. When losing communication with the operator, the robot must independently, using the received and remembered information about outdoor world, Go back back to the original position.

Managing an autonomous robot from the operator is becoming a new character. This is no longer direct movement control, but setting tasks. Since the conditions for performing tasks are not always observed, the Office acquires the nature of the dialogue between the person and the intelligent control system. The latter takes equal participation in planning operations and decision-making. This kind of robotic systems is called cooperative management systems.

The scope of autonomous robots is very wide. This is a search and neutralization of hazardous facilities, problems of radiation and chemical intelligence, work in the zone of man-made and natural disasters. Such robotic systems are also used in the civilian sphere as a service robotics. Service robots have already appeared and successfully perform the functions of servicing visitors in museums, airports, shops. It is especially important to apply service robots in medical institutions, including as a means of rehabilitation of patients. The telepresence service robots are actively used, allowing you to remotely indoors and move along it, observing the robot video camera.

The practical application of autonomous mobile robots led to the need to simultaneously participate in the fulfillment of the tasks of not one, but groups of interacting robots. The theory of group management of intellectual robots is at the initial stage of its development. Nevertheless, certain results are also obtained in this area.

1. Navigation and moving in space with moving obstacles. The work of a mobile robot in the room is considered, the plan of which is unknown in advance. Indoors are available as static obstacles (walls,

tables, chairs) and mobile (people, other robots). The mobile robot is equipped with a scanning laser rangefinder, which receives the scan of the relief of the surrounding objects in the plane parallel to the underlying surface. It is necessary to determine the position of the mobile robot in real time in the room coordinates system (localization task), as well as build a map of this room that displays relief of walls and fixed objects. This kind of management system is known as SLAM systems (System of Localization and Mapping).

The functional structure of the mobile robot navigation system is shown in Fig. The 1 feature of the proposed structure is independence from the type of chassis of a mobile robot, as well as on the presence and type of sensors of the ODO-metry, which allows the use of the developed navigation system on all types of mobile robots operating in the room.

To fulfill its tasks, the robot should move along a given route and at the same time observe security measures, including if there are moving objects in the working area. Thus, the robot moves autonomously with the help of the navigation system, while the operator only performs the function of setting the problem. The semi-automatic mode is not excluded, for example, the telepresence mode, in which the operator's task is significantly simplified.

At the first stage of the navigation system, the task of filtering the scan is solved by removing false measurements using a special filter. To solve the problem of analyzing the model of the working environment, the method of normal distributions was first studied (NDT - Normal Distribution Transform). In this case, the room map is divided into cells, each of which contains not the points themselves, but the parameters of the normal distribution of all points falling inside. Resolving the task of minimizing the function of the mutual correlation of the scan and card, one can determine the position of the robot from which the current scan was obtained

Fig. 1. Functional diagram of the navigation system of the mobile robot

Analysis of this method showed that it significantly limits the speed of movement of the mobile robot, since all calculations are performed real-time. Therefore, a new method was proposed in which the scan information sensors obtained using the system is converted into a grid function. At the same time, each point of the scan is converted into some continuous function, then they are combined using the selected superposition principle and are superimposed on the mesh card, thereby forming a grid function. The method of grid functions, as well as the method of normal distributions, is based on a comparison of the scan and the received card with a mutual correlation function. To minimize this function, a modified Newton method was used. After solving the SLAM problem, the position of the robot on the card can be calculated using the conversion from the coordinate system of the laser rangefinder to the coordinate system of the mobile robot. Comparative analysis The two considered localization methods showed that the advantage of a new method of grid functions is an extended region of convergence, which makes it possible to noticeably increase the speed of movement of the mobile robot.

The specifics of the mobile robot management in the dynamic environment is that the movement of obstacles cannot be calculated in advance. To avoid collisions with moving obstacles, it is necessary to know their position and predict the trajectory of their movement. Then you can move along a planned trajectory, deviating from it at the right moment to make maneuver and drive around the obstacle. A robot control algorithm is proposed in a dynamic environment based on tracing moving obstacles. At the first stage, the task of planning a route along the room constructed by grid functions is solved. This uses a well-known algorithm A *. Further, the task of tracing moving obstacles is solved - determining the current vector of the obstacle status at each time, synchronized to obtain a new scan. To build a list of obstacles, it first takes classification and clustering of scan points. Clustering in this case is carried out in the Euclidean distance between the scan points. The threshold value is calculated based on the distance to the point and the angular resolution of the laser rangefinder.

Each movable object is processed in order to obtain the corresponding parameters of the dimensional circle. After finding a dimensional circle, it is possible to obtain an estimate of the full vector of the status of a movable object (Fig. 2).

The developed algorithm predicts the new position of this object using the kinematics equations (it is assumed that the observed object is a solid body moving along the plane) and the previous vector of its state. Then, using the obtained prediction and parameters of the overall circle, is determined (& + 1) -th estimate of the position of the object:

Fig. 2. To determine the position of the moving object

k + 1 \u003d K ■ HCC + (1 - K) ■ HCR \u003d

MC + 1 \u003d K ■ Casts + (1 - K) ■ Ukr

wHERE KHKI, TILES - the measured position of the object, the CCR, the UKR is the predicted position of the object, K is the filtering coefficient.

Then the length of movement and direction is determined, along which it was performed:

hC + 1 - HC) FKI \u003d 2Ags £

(MC + 1 -Uk u

V HC + 1 - HCCs also calculates the rate of exchange angle:

FC + 1 \u003d KF ■ FKI + (1 - CF) ■ FCR speeds are determined by evaluating the derivative of the length traveled and from the increment of the course angle:

The tasks of following the planned route and bypass moving obstacles are solved together, since when moving along a given trajectory, mobile objects may be near this trajectory. In this case, you need to make a crawler maneuver. Further motion, regardless of the previous state, should be optimal in the selected criterion.

To confirm the proposed management method, an experiment was supplied, including building a room map, determining the accuracy of localization and assessing the quality of controlling the robot in the presence of dynamic obstacles (Fig. 3).

Fig. 3. Conducting experiments on navigation Mobile Robot

The measured accuracy of localization was 0.68 ± 0.45 ° at an angle of rotation and 0.4 ± 0.8 cm on the position, which is a good result exceeding the existing analogues. The quality quality indicators amounted to the length of the path, the time of performing the operation and the distance to the obstacle, respectively: lpath \u003d 2.4 ± 0.6 m, tray \u003d 6.8 ± 1.3 s, Dmah \u003d 0.5 ± 0.2 m.

The developed navigation system has found its application in a real service robot built by Neurobotix to patrol the premises in order to find persons with a changed psycho-emotional state. Region of its use: train stations, airports, shopping centers and other places of massacre of people.

2. The task of the automatic return of the robot. Note that, despite the greater degree of autonomy of the intellectual mobile robot, it should still be controlled by the operator who puts current tasks in accordance with the information received. In addition, in most tasks, mobile robots are directly controlled in remote mode by the operator. With loss of communication, impairment of visibility and other cases, when semi-automatic control cannot be used, the problem of automatic return of the robot to the operator arises. If the main tasks are solved in remote mode, then it is not always appropriate to apply expensive laser rangefinder to solve this problem by the SLAM problem. Therefore, in extreme situations, the alternative to these methods is visual odometry - the method of estimating the linear and angular displacement of the robot by analyzing the sequence of images taken by the camera installed on it.

In tab. 1 is a comparison of the main methods used today to solve the tasks of navigating mobile robots. Here are the advantages of "+" and the disadvantages of "-" considered methods of navigation when solving the problem of returning the robot. This comparison shows the feasibility of applying the method of visual odometry.

Table 1

Comparison of ways to navigate autonomous mobile robots

S% on Slam Lighthouses Wheel-Inertz. ODO-METER VISUALN. Odometry

Do not accumulate + + + - - - -

High accuracy on short trajectories - - +/- + +

Works in a non-deterministic medium + - + + +

Does not reduce accuracy indoors - - + + +

The size of the territory is not limited. + - - + +

Invariant to slipping wheels + + + - +

Does not require stationary objects + - - + +

Invariant to changes in Wednesday + + - + +

In fig. 4 shows the sequence of the steps of the visual odometer. For its work, there are periodically entered pairs of images obtained by two cameras located on board the robot, and for each of the procedures listed in Fig.4. After alignment of images on the left image, special points are distinguished, which can be stable from others, for example, objects of objects, stains, sharp brightness differences, etc. The right image contains the corresponding points. The calculation of the spatial coordinates of the singular points is performed when solving the triangulation task using the difference in the position of the images of the same point with a two cameras. To calculate the displacement of the robot, the change in the position of the singular points over time is monitored. A lucas and Keneda algorithm and Hirschmüller filtering are used.

Fig. 4. Sequence of the steps of the visual odometer

Calculation of the movement of the robot - statistical task. The accuracy of its solution is achieved due to the number of points. In fig. 5 shows the movement of the coordinate system of the robot during the shift of frames. The coordinate system 0 ° X0U0 20 is fixed. 0Shsus2c - the position of the coordinate system of the robot at the current time. 0 P chrur2r - at some early time. - fixed special points of space. Let the coordinates of the special point of space in the coordinate system of the robot at the current time of the time of the SG \u003d (X with the UUS 2C) T, the coordinates of the same point of space at some previous point of time p1 \u003d (XP Ur 2p) T, n dots were found. Changing the position of the coordinate system of the robot will be described in the form:

C (\u003d yar1 + t, 1 \u003d mg,

(G11 g12 g13 \\

r21 g2 2 g2 s i - matrix

company. The task is to obtain an estimate T and and when solving the resulting (redefined) system of linear algebraic equations by the method of smallest squares. The more coordinates are fixed, the less solution is sensitive to the error in determining the coordinates of special points.

Fig. 5. Moving the coordinate system of the robot

The current coordinates of the robot are estimated at the notes of displacements T and R. When the singular points are disappeared from the field of vision, they are excluded and replaced with new ones, which makes it possible to measure the coordinates in the changing medium. Determination of the coordinates of the robot allows you to build a trajectory of automatic return to the operator using third degree spline.

The control system scheme that provides a description of the trajectory according to visual odometry and the automatic return is shown in Fig. 6. In the usual mode "Manual" operator controls the robot using the joystick J, setting the linear and angular velocity V ** and W **. These signals are accepted through the RF transceiver and are entered into the drive control system D where they are converted into angular velocities of wheels Q1 and Q "2. The Visual VO Odometer calculates the linear and angular velocity of the robot and, the number of which allows you to estimate its current X, Y coordinates and course a. NVG navigation unit saves a robot trajectory, removing hinges and noise. When losing communication, the RF block will switch the system into the "auto" mode, in which the robot returns to the saved path automatically. For this, the NVG block displays the X * and YE coordinates of the PF. . The robot is moving at this point using the visual odometer data as feedback on the position.

An experimental study of the proposed system of automatic return, installed on a mobile robot, was carried out in real conditions. At the first stage, the operator with the help of the joystick showered the robot to the end point along the arbitrary trance-thoria, which recorded the proposed algorithm.

Fig. 6. Mobile Robot Management System Scheme

When the robot reached the endpoint, the loss of communication was simulated, and the operator switched the control system into automatic mode. At the second stage, the robot was automatically returned to the operator (at the starting point) on the preserved description of the trajectory.

In each race after the return of the robot, the coordinates of the stop point were measured and the average quadratic error of the robot output was measured at the starting point relative to the length of the trajectory. In total, about 100 races on various trajectories were performed for checking. Experimental studies were carried out indoors, on the street and in the park. They showed that the best results can be obtained by complexing the data of a visual odometer with other means of odometry - wheel odometry and with data of an inertial meter of angular velocity. The routine return error in the starting point was not more than 3%. Note that the experiments were carried out in unfavorable for visual odometry

via: at dusk and darkness, on monophonic underlying surfaces. Nevertheless, in these conditions, the use of visual odometry has significantly improved the accuracy of the navigation system.

3. Fuzzy model of the current situation. The management of attional mobile robots by the operator acquires the nature of the formulation of tasks and dialogue accompanying their implementation. At the same time, a spatial-temporal relationship should be used from the point of view of a person, which greatly facilitates the task of controlling the robot. To organize the Mobile Robot dialogue management, it is advisable to use fuzzy logic methods. These methods proved very effective for describing the external world and the current situation with the help of linguistic variables. The development of this approach, in relation to the task of managing robots, is to find a "natural" in the same sense to assess the situation, and to adopt in advance indefinite conditions, which determines the behavior of the robot.

The description of the external world of the robot includes both the description of objects of interest to perform a given operation and spatial relations between the objects of the world, including the robot itself. To describe the spatial relationship between objects of the working scene, extensional and intente fuzzy relations are used. The first is the relationship of the situation and orientation of objects. For example, "Object A1 is far away, and ahead of the Oblast A2." Intente relations include such relationships, how to contact; be be out of be in the center, etc. From elementary spatial binary relationships, using formal rules for conjunction and dysuunction, get other relationships in practice.

The current situation, which includes M objects, including the robot controlled, is described by the Binary Frames system (<объект m>, <отношение>, <объект п>), M, I \u003d 1.2, ..., m. If fuzzy binary relationships between all objects are installed in advance, which can be observed by a robot during the movement, then we will receive a fuzzy semantic network, or a fuzzy card. Using such a map, you can, in particular, navigate the robot on the observed references, i.e. According to objects whose position was known in advance. The image of the current situation may include other fuzzy signs, in addition to spatial. For example, a mobile robot designed to protect the room from a fire may have sensors of temperature, humidity, air composition (presence of harmful substances or smoke), acoustic sensors. The combination of this data determines the direction of movement to the fire source.

An important feature of the management task using a fuzzy working scene model by a mobile robot equipped with a technical vision system is that during the movement, the scale of the image perceived by the cameras installed on the robot changes. It changes depending on the distance to the obstacle and the course angle. This effect necessitates the introduction of two-dimensional accessories to determine the current position of the robot on the plane. For example, the function of orientation affiliation (for course angle) depends on the range D. This feature of the system of technical violation of the autonomous mobile robot corresponds to the law of the spatial perspective inherent in the natural vision of man.

Since the outside world is continuously changing both due to the movement of the observed objects and due to the movement of the robot itself, then the description of the situation varies in time. This circumstance requires accounting in the general case not only spatial, but also temporary relations in the external world, such as

"Being at the same time", "be earlier", "follow for". Such relationships have to be used, in particular, when managing mobile robots moving in space, containing other moving objects. They allow you to provide automatic support of moving objects, or avoid collisions with them.

In the general case, the situation is determined by frame, which the names of the objects of the outside world, the natural relationship between objects, both spatial and temporary, as well as other signs characterizing the situation. A comparison of the observed situation with one of the reference situations contained in the knowledge base is carried out using certain criteria for fuzzy's proximity of situations, so that the assessment assessment thus obtained by a robot represents nothing more than a formalized and averaged assessment of a similar situation by man.

The set of designations (names) of the specified objects in the space of the working scene and fuzzy relations between them is a dictionary of the formal language description language. Using the introduced D.A. Pospel the terminology, you can call the language used to describe situations of formal logical relations, situational language. In a situational language, a dialogue between the robot and man can be organized when analyzing the situation. The dialogue organization of the interface allows the robot to form requests to the operator in the absence necessary information To build a model of the outside world.

4. Mobile robot dialogue. By describing the current situation in the language of linguistic variables and fuzzy relations, you can set the behavior of an autonomous robot in an external, not fully defined medium using fuzzy product rules. You can a priori allocate stereotypes of behavior defined by the current situation. They are sometimes considered as an analogue of the conditionally reflex behavior "Stimulus-reaction". These behavioral stereotypes have the type of product rules: "If the situation is Si, then tactics T." Under the tactics, we understand the combination of the rules of behavioral expressed using linguistic variables and the determined goal. These rules put in line with a typical situation of a predetermined model movement of the robot. For example: "If the goal is far, then move quickly," if the goal is close, then move slowly. " It is assumed that typical situations can be in advance in a fuzzy robot knowledge base, using the experience of the human operator. Using this database, you can make a set of rules of behavior (tactics) corresponding to the persecution of a new object, output to a specific point specified on electronic map, aisle in the doorway, bypassing a sudden obstacle to the goal, accompanied by moving objects, etc. The presence of a tactic obtained by a robot in advance, or in the learning process, significantly simplifies the task of the operator, eliminating it from managing the robot when solving routine tasks. At the same time, the tactics themselves may be laid into the robot knowledge base by teaching a neuro-fuzzy network, which is the basis of a fuzzy controller.

The purpose of managing a mobile robot, the control system of which contains a fuzzy controller, is formalized as a system of fuzzy decision rules. To make a decision with a fuzzy mobile robot controller, one of the well-known fuzzy output schemes can be used. The Mamdani scheme containing both the phase procepter (bringing to the fuzziness) of the sensors determining the relative position of the robot and goals and the defazzzynation procedure, i.e. Bringing to the definition of managed commands obtained in fuzzy form, which is necessary to generate control signals.

In the general case, the tactics of the behavior of the robot is determined by the framework of the problem, which can be submitted in the following form:<текущая ситуация S1> <объект управления а0> <имя операции> Related objects J\u003e< условия выполнимости операции >. The default management object is the mobile robot itself, the capabilities of which are in the database. These capabilities (dimensions, weight, power of drivers, speed, maneuverability, etc.) determine the conditions for the feasibility of the operation, taking into account the current situation and the properties of the working medium (relief, the grip of the wheels with the soil, carrying the properties of the soil, the nature of the obstacles). Performance conditions operations may include and checking the postlands that must be completed after the operation is completed. Perhaps the conditions for the feasibility of the operation will require the implementation of special search movements that we refer to "cognitive operations" and which should also be contained in the knowledge base of the robot.

The operator command typically includes only two elements:<имя операции > <сопутствующие объекты>. For example,<преодолеть > <порог П>. In this case, the formalization of the current situation and checking the conditions for the operation must be carried out by the system itself.

The operator can inform the robot only the final goal of movement in space with a partially known structure. In this case, there is a problem of autonomous planning of motion, requiring special consideration. The solution to such a task ensures the laying of the trajectory in space with known obstacles, which has the best in one way or another quality indicators and its redevelopment when unforeseen obstacles appear. Planning is accompanied by a dialogue with the operator when ambiguities appear or, if necessary, clarify the goal. If the goal is incorrect, or the scheduler did not reveal the realizable sequence of movements leading to the target, then the request occurs to the operator, requiring clarification of the goal or the conditions for its achievability.

Organization dialogue system Management involves creating a speech interface that includes the recognition module and a linguistic analyzer. The first module is a device for converting speech signals and their interpretation as separate words or phrases. Linguistic analyzer performs syntactic and semantic analysis Spellings, as a result of which the frame slots should be filled to describe actions.

The use of a dialogue management system significantly simplifies the problem of controlling the robot, since in this case there is practically no availability of special skills from the operator. Therefore, it can be assumed that the dialogue management will be widely distributed in the near future when managing service and special robots of various purposes.

5. Management of a group of autonomous robots. The practical application of intelligent mobile robots in the areas associated with a person for a person led to the conclusion about the feasibility of managing a group of robots, jointly decisive the overall task. Characteristic examples of such tasks are intelligence, rescue and liquidation operations during accidents on various chemical and radiation-hazardous facilities. At the same time, the problem of organizing the agreed work of autonomous interacting robots, forming a multi-age-age robototechnical system (MRS).

Preliminary analysis showed that in order to solve complex problems in the face of uncertainty, the organization of a distributed hybrid type system, which is a combination of centralized and

decentralized systems. Here, the Collective Management is performed by exchanging information between agents, and data processing is carried out by the control center.

The system includes a managing center, a team of robot agents who have different "specialization" and auxiliary equipment. The functions of the human operator include monitoring the system status and decision-making in difficult situations. The Managing Center forms the team management commands in general and processes the information obtained by robots during the task execution process. Robots, depending on their functionalityCan perform reconnaissance, search or technological operations. In accordance with the task of the control center, the robots using messaging are combined into groups and distribute problems with each other, taking into account the specified optimal criterion.

In fig. 7 shows the structural scheme of the developed hierarchical control system of the MRS, built according to the modular principle. The control center forms the first level of the control system contains a block of formation of commands that is responsible for developing a command sequence and transmitting them to robots, and a data processing unit where information comes from agents-robots and auxiliary information and measuring equipment.

The second level is responsible for the global planning of the MRTC actions: the decomposition of tasks solved by MRTs, on simpler tasks and the distribution of them between robots. The result of the work of the corresponding algorithms is to determine the coordinates of the target point for each robot.

The third level ensures the formation of trajectories of moving robots using local planning algorithms based on the analogue of the potential method. To form a trajectory of motion, it is necessary to determine the current coordinates of robots in the workspace, this task is solved by a combined navigation system.

Fig. 7. Structure of the MRC Management System

The fourth level (executive) is directly interaction with all elements of the MRTC and the external environment, it is responsible for the production of control signals for robot drives, navigation sensors, special and auxiliary equipment, supports the operation of the sensory system.

Solving global planning tasks is necessary when implementing both centralized and multiagent management. In the case of centralized management of the task in the robot team, the control center distributes in accordance with the specified optimality criterion (as a rule, this is the time for solving a global task).

The easiest way to solve the task of centralized management is a generation method - to apply in most cases it is inappropriate, since it requires a significant solution time. Evolutionary methods (genetic algorithms) also require considerable time due to computational complexity. In this regard, to control real-time time, the robots group was developed, implemented and studied the simpler - "price" algorithm. In this case, the control center for each robot forms a price array (SU), where i \u003d 1.2, ..., and the number of robots, y \u003d 1,2, ..., T - task numbers. This array contains the estimated robot costs (for example, time or energy consumption) to perform each task. The managing center must distribute the tasks in the robots group in such a way that the corresponding values \u200b\u200bof "prices" sought to minimally (Fig. 8).

Fig. 8. Centralized Fig. 9. Multi-agent distribution

distribution of tasks in the group of tasks in the group

With a large number of robots included in the group, centralized management has a number of disadvantages, among which the low reliability of the system, a large load on the radio channels and a high consumption of computational resources of the control center. In addition, the centralized approach is ineffective with the rapid change in the parameters of the working area. These shortcomings are deprived of the second approach to the distribution of tasks in the robot group - multi-agent management, when robots in the negotiation process independently distribute the tasks among themselves, also guided by the specified optimality criterion (Fig. 9).

Such a distribution of tasks best provides a "auction" bar negotiation model based on the exchange of information between individual agents. At the auction, some resources, such as energy resources necessary to agents to achieve the goal, "are put up for sale." Since resources are limited, the agents compete with each other in the process of "auction". In the distribution of tasks in the team of robots, the tasks themselves act as resources. The feasibility of "shopping", i.e. Appointments The robot of a particular problem is assessed by the specified optimality criterion.

At each stage of the auction, the robots agents form their own price arrays, and one of the team robots becomes the leader. The leader chooses from the responding agents of those that offer the smallest price, and sends them

command perform task. At the next stage, the leader becomes another, free from the task, the agent. The auction is carried out until all tasks are distributed. Multi-agent management provides the ability to self-organization of the system and increases the reliability of its operation. A comparison of multi-agent methods has shown that the task distribution algorithm developed on the basis of the auction negotiation model has a significantly smaller computational complexity (at 0.12 * n times, with n \u003e\u003e 8) than the famous multi-gene-generating method of collective improvement plan.

To implement a multiagent distribution of tasks, each robot agent should be able to exchange information with all other robots that are part of the collective. In real conditions, it is not always possible to fulfill these requirements, since within the working zone there may be various obstacles and interference violating information exchange. The solution to this problem is to split the team of robots into individual groups, each of which has a leader agent, capable of sharing data with all other group agents. At the same time, the number of leaders should strive to a minimum. As part of the developed multiagent algorithm for the formation of autonomous groups of robots, each agent using a request message determines its neighbors - robots with which can exchange information, counts their number and sends this value to their neighbors. Then each robot chooses from the list of its neighbors leader-agent with the highest number of neighbors. As a result, the robot team is divided into groups having leaders. After that, inside each formed group, you can use the previously multi-agent task distribution algorithm.

Thus, the result of global planning is to split the team of robots into groups, within which each robot has an appropriate task. All tasks solved by robots in groups are reduced either to the uniform distribution of robots in the working area, or to the movement of robots to the specified target points and perform any operations at these points, so the control system must ensure the formation of robot movement trajectories, that is, decide Local planning tasks. The corresponding algorithms are implemented at the third level of the control system and are basic for the formation block of the trajectories of algorithms for the formation of the trajectories is the kinematic analogue of the known method of "potentials".

The implementation of this method ensures a uniform distribution of robots in the working area during its survey to identify emergencies, as well as moving robots to the required target points to perform technological operations. In both cases, robots in the process of movement should avoid collisions with static obstacles and with other robots.

The purpose of the territory survey is to build a map of the working area. This can be both a physical map of the terrain and a special cartographer obtained from the results of information processing collected by the team of robots, for example, information about radioactive infection. To collect these robots are uniformly distributed within the working area and measure the values \u200b\u200bof the required parameters. Measurement results are received in the control center database and are applied to a special area of \u200b\u200bthe terrain. The resulting cartographer represents a certain surface of r \u003d / (x, y), where the pair of x, y coordinates of the point, G is the level of chemical or radiation infection. This surface must be provided to the operator in a convenient form. Most

a simple and visual way to build a two-dimensional color (gradient) cartographer. The cartogram, in addition to the information for the operator, can be used to form secure routes of the evacuation of people. Algorithm for building secure routes with a guaranteed degree of safety P *, i.e. With the value of the specified FK parameter, which will not be exceeded at any point of the route found, is based on the Bellman-Ford algorithm for the shortest path in the suspended column. Figure 10 shows the result of the operation of the algorithm for constructing the shortest secure route on the gradient cartogram with a guaranteed degree of safety P \u003d 1.5.

Fig. 10. The shortest secure route on the cartogram

To verify the proposed methods, a software package has been developed that implements the algorithms of centralized and multiagent management of the team of robots, algorithms for the formation of the trajectories of the movement of robots, as well as methods of processing information collected by robots. The result of the processing of the source data is the software model of the physical map of the area with the specified obstacles, which is displayed in the animation window (Fig. 11).

Ammaiyim window

* * ♦. "-" __ "J 1

, ® h ~ 1 "- g! "- | |

Fig. 11. Marking of safe routes and "neutralization" of foci of infection

The foci and intensity of chemical infection are displayed in the corresponding window in the form of a cartogram. Further, the operator is given the ability to serial selection of control commands and, thus, starting the modeling process of the corresponding MRTS operation modes. On the team "Examination", robots are evenly distributed in the working area and measure the

the centers of poisoning substances. Points in which concentration exceeds permissible levelmarked on the map and, according to the measurement results, the contamination card is adjusted. According to the "Search for Secure Route" command, the routes of staff evacuation are built automatically, and on the team "Marking and Neutralization", robots are broken into groups in accordance with their specialization. Part of the robots moves to the foci of infection for their neutralization, and other robots are placed along the evacuation routes as markers (in Fig. 11 they are equipped with flags).

The developed multi-age-age robotic system can be used both in the case of man-made accidents on chemically hazardous sites, under conditions of radiation or biological infection (using special equipment) and when eliminating the consequences of various natural disasters.

Conclusion. The development of robotics enters into a new stage when we proceed to the management of mobile and manipulative robototechnical devices, we move to the management of a cooperative type, in which the robot becomes a full-fledged participant in the management partner of the operator. It is significantly simplified by the task of the operator, which is practically not required pre-training. However, the robotic system itself becomes more complicated, which is now a high degree of autonomy and possessing the capabilities that belong to artificial intelligence. Due to this, it becomes possible to solve with the help of robotic devices a significantly wider range of tasks than before. Problems are now resting, on the one hand, in the ability computer equipmentwhich should evaluate the current situation and manage mobile robots real-time taking into account sufficiently high speeds. And on the other hand, in the psychophysiological capabilities of the human operator, accompanied by the autonomous activity of mobile robots in the outside world. If the first problem is gradually solved in the course of technical progress, the second, due to the limited personality of the person, requires the continuous development of the "Robot person" interface, taking into account his psychological restrictions.

Bibliographic list

1. Kristensen S, Horstmann S., Klandt J., Lohner F., And Stopp A. Human-Friendly Interaction for Learning and Cooperation // Proceedings of the 2001 Ieee International Conference on Robotics and Automation, Seoul, Korea, 2001. IEEE . - P. 2590-2595.

2. Gerasimov V.N., Mikhailov B.B. Solving the problem of controlling the movement of a mobile robot in the presence of dynamic obstacles // Bulletin MGTU them. AD Bauman. Instrument making. Special issue "Robotic systems". - 2012. - № 6. - P. 83-92.

3. Ulas C., Temeltas H. Multi-Layered Normal Distribution Transform for Fast and Long Range Matching // Journal of Intelligrnt & Robotic Systems. - 2013. - Vol. 71 (1). - P. 85-108.

4. Gerasimov V.N. SLAM algorithm based on correlation function // Extreme Robotics: Collection of reports of the All-Russian Scientific and Technical Conference.

St. Petersburg: Publishing House "Polytechnic-Service", 2015. - P. 126-133.

5. Gerasimov V.N. To the issue of moving the movement of a mobile robot in a dynamic environment // Robotics and technical cybernetics. - 2014. - № 1 (2).

6. Nister D., Naroditsky O., Bergen J. Visual Odometry for Ground Vehicle Applications // Journal of Field Robotics. - 2006. - Vol. 23 (1). - P. 3-20.

7. Nahrikov E.A., Mikhailov B.B. Using a visual odometer data for autonomous return of a mobile robot in a medium without fixed reference points // Extreme Robotics: Collection of reports of the All-Russian Scientific and Technical Conference. - SPb.: Publishing House "Polytechnic-Service", 2015. - P. 356-361.

8. Motorerics E.A. The algorithm for describing the trajectory of a mobile robot according to the visual odometer for automatic return to the operator // Science and Education. MSTU them. AD Bauman. - 2014. - № 12. - P. 705-715.

9. Motternikov E.A., Mikhailov B.B. Visual odometer // Bulletin MGTU them. AD Bauman. Ser. Instrument making. Special issue "Robotic systems". - 2012. - № 6.

10. Kradrashina E.Yu., Litvintseva L.V., Pospelov D.A. Representation of knowledge about time and space in intelligent systems. - M.: Science, 1989. - 328 p.

11. Yushchenko A.S. Routing Mobile Robot Movement in Uncertainty Conditions // Mechatronics, Automation and Management. - 2004. - № 1.

12. Yushchenko A.S., Tachkov A.A. Integrated firewall control system reconnaissance robot // Bulletin MSTU them. AD Bauman. Instrument making. Special issue "Robotic systems". - 2012. - № 6. - P. 106-11.

13. Collars S.A., Yermishin K.V. Intelligent Service Mobile Robot Management System // Bulletin MGTU them. AD Bauman. Instrument making. Special issue "Robotic systems". - 2012. - № 6. - P. 285-289

14. Melikhov A.N., Bernstein L.S., Korovin S.Ya. Situational consistent systems with fuzzy logic. - M.: Science: Fizmatlit, 1990. - 271 p.

15. Yushchenko A.S. Dialogue management robots based on fuzzy logic // Extreme Robotics: Collection of reports of the All-Russian Scientific and Technical Conference. - St. Petersburg: Publishing House "Polytechnic-Service", 2015. - P. 143-146.

16. Yushchenko A.S. Intellectual planning in the activities of robots // Mechatronics, automation, management. - 2005. - № 3. - P. 5-18.

17. Jonin A.A. Algorithm of learning manager of the dialogue of the speech dialogue management system robot // Integrated models and soft calculations in artificial intelligence: a collection of scientific papers international Conference. - M.: Fizmatlit, 2011. - P. 395-406.

18. Nazarova A.V., Ryzhova TP Methods and algorithms of multi-agent management by the robot-technical system // Most Utra, MSTU. AD Bauman. Instrument making. - 2012.

- № 6. - P. 93-105.

19. Jennings N., Paratin P., Jonson M. Using Intelligent Agents to Manage Business Processes // The Practical Application Of Intelligent Agents and Multi-Agent Technology: Proceedings of the First Intern. Conference. London (UK). - 1996. - P. 345-376.

20. Kapustyan S.G. The algorithm for collective improvement of the plan in solving the tasks of the distribution of goals in the robot group // piece intelligence. - 2005. - № 3. - P. 463-474.

21. Darinsev O.V. Microbot collective control system // piece intelligence.

2006. - № 4. - P. 391-399.

1. Kristensen S, Horstmann S., Klandt J., Lohner F., And Stopp A. Human-Friendly Interaction for Learning and Cooperation, Proceedings of the 2001 Ieee International Conference on Robotics and Automation, Seoul, Korea, 2001. IEEE, PP. 2590-2595.

2. Gerasimov V.N., Mikhaylov B.B. Reshenie Zadachi Upravleniya Dvizheniem Mobil "Nogo Robota PRI Nalichii Dinamicheskikh Prepyatstviy, Vestnik MgTU Im. N.E. Baumana. Priborostroenie. SpetSvypusk" Robototekhnicheskie Sistemy ", 2012, No. 6, pp. 83-92.

3. ULAS C., Temeltas H. Multi-Layered Normal Distribution Transform for Fast and Long Range Matching, Journal of Intelligent & Robotic Systems, 2013, Vol. 71 (1), pp. 85-108.

4. Gerasimov v.N. Algoritm Slam Na Osnove Korrelyatsionnoy Funktsii, Ekstremal "Naya Robototekhnika: Sbornik Dokladov Vserossiyskoy Nauchno-Tekhnicheskoy Konferentsii. St. Petersburg: IZD-VO" Politekhnika-Servis ", 2015, PP. 126-133.

5. Gerasimov v.N. K Voprosu Upravleniya Dvizheniem Mobil "Nogo Robota V Dinamicheskoy Srede, RoboTotekhnika I Tekhnicheskaya Kibernetika, 2014, No. 1 (2), PP. 44-51.

6. Nist "ER D., Naroditsky O., Bergen J. Visual Odometry for Ground Vehicle Applications, Journal of Field Robotics, 2006, Vol. 23 (1), pp. 3-20.

7. Devyaterikov E.A., Mikhaylov B.B. Ispol "zovanie dannykh vizual" nogo odometra dlya avtonomnogo vozvrashcheniya mobil "nogo robota v srede bez fiksirovannykh tochek otscheta, Ekstremal" naya robototekhnika: Sbornik dokladov vserossiyskoy nauchno-tekhnicheskoy konferentsii. St. Petersburg: IZD-VO "Politekhnika-Servis", 2015, PP. 356-361.

8. Devyaterikov E.A. ALGORITM OPISANIYA TRAEKTORII MOBIL "NOGO ROBOTA PO DANNYM VIZUAL" NOGO ODOMETRA DLYA AVTOMATICHESKOGO VOZVRASHCHENIYA K OPERATORU, NAUKA I OBRAZOVANIE. Mgtu im. N.E. Baumana, 2014, No. 12, pp. 705-715.

9. Devyaterikov E.A., Mikhaylov B.B. Vizual "Nyy Odometr, vestnikmgtu im. N.E. Baumana. Priborostroenie. SpetSvypusk" Robototekhnicheskie Sistemy ", 2012, No. 6, pp. 68-82.

10. Kandrashina E.Yu., Litvintseva L.V., Pospelov D.A. Predstavlenie Znaniy O Vremeni I Prostranstve V Intellektual "Nykh Sistemakh. Moscow: Nauka, 1989, 328 p.

11. Yushchenko A.S. Marshrutizatsiya Dvizheniya Mobil "Nogo Robota V USLOVIYAKH NEOPREDELENNOSTI, MEKHATRONIKA, AVTOMATIZATSIYA I UPRAVLENIE, 2004, No. 1.

12. Yushchenko A.S., Tachkov A.A. INTEGRIROVANNAYA SISTEMA UPRAVLENIYA POZHARNYM RAZVEDYVATEL "NYM ROBOTOM, VESTNIK MGTU Im. N.E. Baumana. Priborostroenie. SpetSvypusk" Robototekhnicheskie Sistemy ", 2012, No. 6, pp. 106-11.

13. Vorotnikov S.A., Ermishin K.V. Intellektual "Naya Sistema Upravleniya Servisnym Mobil" NYM Robotom, Vestnik MgTU Im. N.E. Baumana. Priborostroenie. Spetsvypusk "Robototekhnicheskie Sistemy", 2012, No. 6, pp. 285-289.

14. Melikhov A.N., Bernshteyn L.S., Korovin S.Ya. SituatsionNye Sovetuyushchie Sistemy S Nechettokoy Logikoy. MOSCOW: Nauka: Fizmatlit, 1990, 271 p.

15. Yushchenkoa.s. Dialogovoe upravlenie robotami na osnove nechetkoy logiki, Ekstremal "naya robototekhnika: Sbornik dokladov vserossiyskoy nauchno-tekhnicheskoy konferentsii St. Petersburg:. Izd-vo« Politekhnika-servis », 2015, pp 143-146..

16. Yushchenko A.S. Intellektual "NOE PLANIROVANIE V DEYATEL" NOSTI ROBOTOV, MEKHATRONIKA, AVTOMATIZATSIYA, UPRAVLENIE, 2005, NO. 3, pp. 5-18.

17. Zhonin A.A. Algoritm Obucheniya Menedzhera Dialoga Rechevoy Dialogovoy Sistemy Upravleniya Robotom, Integrirovannye Modeli I Myagkie Vychisleniya V Iskusstvennom Intellekte: Sbornik Nauchnykh Trudov Mezhdunarodnoy Konferentsii. MOSCOW: Fizmatlit, 2011, PP. 395-406.

18. Nazarova A.V., Ryzhova T.P. Metody I Algoritmy Mul "Tiagentnogo Upravleniya Robototekhnicheskoy Sistemoy, Vestnik MgTU Im. N.E. Baumana. Priborostroenie, 2012, No. 6, pp. 93-105.

19. Jennings N., Paratin P., Jonson M. Using Intelligent Agents To Manage Business Processes, The Practical Application Of Intelligent Agents and Multi-Agent Technology: Proceedings of the First Intern. Conference. London (UK), 1996, pp. 345-376.

20. Kapustyan S.G. Algoritm Kollektivnogo Uluchsheniya Plana Pri Reshenii Zadach Raspredleniya Tseley V Gruppe Robotov, Shtuchnyy Intellect, 2005, No. 3, pp. 463-474.

21. Darintsev O.V. Sistema Upravleniya Kollektivom Mikroorobotov, Shtuchnyy Intellect, 2006, No. 4, pp. 391-399.

Yushchenko Arkady Semenovich - Moscow State Technical University. AD Bauman (MSTU them. N.E, Bauman); E-mail: [Email Protected]; 105037, Moscow, Izmailovskaya pl., 7; Scientific and Training Center "Robotics" MSTU them. AD Bauman; D.T.N.; Professor.

Mikhailov Boris Borisovich - Scientific and Training Center "Robotics" MGTU them. AD Bauman; k.t.; assistant professor.

Nazarova Anid Vartanovna - Scientific and Training Center "Robotics" MGTU them. AD Bauman, k.t.n.; assistant professor.

Yuschenko Arkady Semenovich - Bauman Moscow State Technical University (BMSTU); E-mail: [Email Protected]; 7, IZMAILOVKAYA SQ., Moscow, 105037, RUSSIA; Scientific-Educational Center "Robototechnika" BMSTU; DR of ENG. SC.; Professor,

Mikhailov Boris Borisovich - Scientific-Educational Center "Robototechnika" BMSstu, Cand. Of ENG. SC.; Associate Professor.

Nazarova Anaid Vartanovna - Scientific-Educational Center "Robototechnika" BMSTU; Cand. Of ENG. SC.; Associate Professor.

UDC 681.511.4 + 004.896: 519.876.5

V.F. Guzik, V.A. Pereverzev, A.O. Poyavchenko, R.V. Saprykin

Principles of constructing an extrapolarizing multidimensional neural network planner of the intellectual system of positional and trajectory management of moving objects *

The principles of constructing an extrapolating multidimensional neural network planner (EMNP) of the intellectual system of positional and trajectory control of moving objects are considered. So here are the results of the study of the upgraded method of neural network planning of the movement of a robotic mobile object in relation to multidimensional space using the bionic principle of surrounding the environment in conditions of uncertainty and the presence of obstacles with the dynamics of movement. As the basic principle of structuring and constructing EMNP, the hierarchical principle of constructing information processing systems was proposed, on the basis of which the hierarchical structure proposed in the article was synthesized in the article by an integrated extrapolizing multidimensional neuro-like network. Such a multidimensional neuropod-like network distinguishes the presence of separate layers intended for various stages of processing the model plan of the medium obtained from the technical vision system of a robotic mobile object in the above conditions mentioned above. The hierarchical structure of a complex multidimensional neuropo-kind network is based on the principles of parametric object-oriented synthesis, the synthesis of spatial plans of weighted features of the position of objects

We are sure that the drones will not be in the very near future. The reason why so many companies are safely promising that it is right around the corner, is that robots give you things when you want this fantastic idea. First, the drones seem good idea, because the flight allows you to quickly reach the place, avoiding obstacles, and people are stuck on this idea for many years, because it would be great if you really could make her work.

So far, it did not work, but this does not mean that the robots supplying things should not happen. And indeed, whether consumers do not care about whether they are delivered by a specially autonomous air drone, if they quickly get their material and should not change their pajamas? Startup, called Starship Technologies, with offices in London and Tallinn, Estonia, announced the creation of an autonomous delivery robot, which promises to do everything that DRONG video can do (and much more), except from the Earth and with the realistic opportunity actually.

How Does Starship Robotic Delivery Work?

Starship Technologies, created by two Skype accomplices, Ahti Heinla (which is the Director General and Technical Director) and Janus Fries, will present Fleams of compact, safe, quiet and most importantly, earthly robots delivery, hoping to open new opportunities for such as shipping firms Packages or grocery stores, as well as create unprecedented amenities and cost savings for individuals. A robot that, as far as I can judge, has no name, can carry the equivalent of two grocery bags (about 10 kilograms) to 5 kilometers from a local hub or shopping store. It moves at the pace, and fully loaded weighs less than 20 kg, which means that the car will be difficult to accidentally cause any harm. Borders and bumps on the road is not a problem, and they seem to do it up and down on some sets of stairs. Integrated software to prevent intrusion and preventing obstacles allows it to be managed mainly independently, but also controlled by operators that may intervene in order to ensure safety at any time.

According to STARSHIP estimates, its robotic supplies will cost 10-15 times smaller than the current alternatives to the delivery of the last mile. According to the company, customers will be able to choose from several short accurate slots to deliver, which means that the goods come at the right time. During delivery, buyers can track the location of the robot in real time through mobile app, And upon arrival, only the owner of the application can unlock the cargo.

We do not say that to make these robots work easily, be sure to: while unmanned aerial vehicles are relatively dangerous, less reliable, no more, more expensive, have a limited load and are currently in legal uncertainty, StarShip robots will have to solve all types of problems which drone drones completely avoid. These problems include searching for ways around roads and sidewalks, navigating next to vehicles and around pedestrians and more direct interaction with people. Moreover, GPS is not accurate enough to keep these robots on sidewalks, so they will need to use the vision to be able to say where it is safe, relying on the localization of the basic card, as the autonomous Google cars do. They will have to understand pedestrian crossings and traffic lights. They will have to listen and react to sirens from emergency cars. And perhaps the most difficult thing, they will have to detect and communicate with unpredictable people.

Nevertheless, we are optimistic about it, because you have already seen a lot of necessary technologies. Robots generally have a lot of practice in safe urban navigation. They know how to avoid obstacles, given enough time and adequate sensor data. There are whole conferences on how to make robots effectively interact with people. The most important thing is that any of these materials does not work, the robot can safely stop and wait infinitely for a person to intervene and help him, with cameras and speakers and microphones providing full telepresence and remote control. Of course, it is worth noting that Aethon, Savioke and other robotic companies have been doing something very similar to this for many years. Delivery of items wandering through hospitals, warehouses and hotels, conjugate with various problems: deliver items, wandering through the streets, but mostly use similar technologies, and the fact that these companies are reliably working, makes us optimistic that StarShip will also be capable of also.

StarShip actively tests prototypes right now, and by next year the company will launch two experimental programs: one in Greenwich in East London and one in the United States.