Statistical methods are widely used for solving of the problems of automatic control of the industrial objects, because they enable us to determine their dynamic characteristics during normal operation of the objects. The statistical correlation method for determination of these dynamic characteristics is based on the solution of an integral equation, which includes the correlation functions R_XX(t) and R_XY(t) of the input Х(t) and of the output Y(t) signals. It allows us to obtain the dynamic characteristics of an object without a disruption of its regular operation mode. However, application of these methods for construction of the mathematical models of the real-life industrial objects presents the following problem. Interferences and noises are imposed upon the useful signal, hindering the calculation of the estimates of their static characteristics. This paper presents one possible option for creation of the alternative methods and technologies for elimination of the error induced by noise during formation of the correlation matrices. The proposed general algorithms allow a reduction of these matrices to the similar matrices of the useful signals. The two presented alternative robust technologies enable one to solve these problems both in the absence of a correlation between the useful signal and the noise, and in the presence of such. The validity of the result is controlled by comparison of the obtained estimates of the elements of matrices by both methods. In many real-life industrial objects we encounter a need to apply the procedure for normalization of their elements. This leads to an additional error, which also leads to a disruption of the adequacy of the results. The authors propose general methods and technologies for elimination of that error.

In this paper the authors analyze a variety of modern discrete control algorithms from the point of view of their computational complexity. The outcome of their analysis revealed high complexity of the major optimization procedures employed in the discrete algorithms of the optimal control. This uncovered disadvantage concerns implementation of the modern control algorithms in the digital real-time controllers of a limited computational power. The article suggests an approach how to adopt the time-consuming algorithms to the low productivity controllers. This approach is based on the idea to divide the computation process into two stages: the first stage includes preventive offline calculation of the optimal control values for a defined set of points in the system's state space; the second stage involves an online interpolation of the previously calculated control values for the current state of the system (only the second stage is realized in a real-time controller). As a result we observe reduction of the requirements to the controller processing power (assuming that the interpolation procedure is less time-consuming than the investigated optimization methods) accompanied with an insignificant decrease of the transient performance of the system. A practical analysis of the obtained results was conducted for the sample optimal control loop based on the numerical procedure of a random search. The random search procedure belongs to the top computation-consuming algorithms and in most cases is not available for implementation in the real-time controllers. Adaptation of this algorithm to the above-described two-steps computational process simplifies the on-line calculations down to the interpolation operation which can be performed by the modern digital controllers of a modest capacity. The findings of this paper were proved by the results of computer simulation.

The authors consider the problem of control of the mobile robots' motion in a leader-follower formation. They use a method, in which the follower position is defined only in the follower-fixed coordinate system and the only information involved is the relative mutual positions of the leader and the follower. Such an approach is more relevant from the viewpoint of applications. For a mobile robot in the formation, the only available information is usually the data from the sensors concerning its relative position with respect to the other robots in the formation. One of the basic methods used for synthesizing of the laws for control of the objects' motion in the formations is the feedback linearization method, although application of this method is limited due to high complexity of the kinematic model and also rather low robustness of this method with respect to the external disturbances. Many researchers dealing with the control of robots in a leader-follower formation use the theory of systems with a variable structure for development of the control laws invariant to the external disturbances. A significant drawback of such control laws is emergence of high-frequency switching in the control signals. Recently specialists have shown an increased interest to the methods of the spatial (functional or program-coordinate) control mainly aimed at a direct solution of the problem of stabilization of the motion over a manifold. A typical feature of the method of the structural synthesis (functional control), also revealed in this work, is the extension of the definition of the relative motion by using an implicit model. Because of the drawbacks of the existing analytical and experimental investigations done by the method of the structural synthesis and also due to certain functional bottlenecks of the existing approaches, the further development of the structural synthesis of the closed control laws on the basis of organizing of a system of motion along a prescribed trajectory in the space of states, is an important task. In this work, the authors use the method of organization of the forced motion along the desired trajectory in the space of states of an object within the problem of control of the motion of a homogeneous formation of the mobile robots with a differential drive. By keeping the required distance between the robots in case of a limited range of interaction of objects in a group by the method of potential functions, it is possible to avoid collisions and obstacles. Numerical experiments confirm the workability of the control system in the presence of the measurement noise and external disturbances.

For an effective use of the mobile robotic platforms, moving in extreme conditions on unpredictable surfaces of different quality, there is a trend for designing of the robots with pneumatic devices with adaptation to such types of surfaces. The manipulators and mobile robots usually employ the vacuum grippers with an ejector system. However, they should have a smaller and more efficient design for application in the miniature mobile robots. The process of movement of robots over the vertical surfaces demands a reliable contact with those surfaces. Selection of the ejector's operation mode is important. In this paper the authors consider a discrete-step way of the robots' movement and techniques of a reliable contact with the surfaces by means of an ejector. The authors also analyze the basic schemes of the adaptive pneumatic grippers of the wall climbing robots under extreme conditions with unknown properties and quality of the surfaces of motion, and present the results of their experimental investigations. Experimental studies were aimed at identifying the properties of adaptation of the robots to the microporous structures, automatic shut-off valves with bellows and feedback. The calculated ratios for assessment of the pressures in cavities in the vacuum area, as well as a constructive scheme of the adaptive pneumatic gripping devices for the mobile robots, were demonstrated. The results are aimed to ensure maneuverability of the robots moving in extreme conditions on the vertical surfaces and ceilings in the unknown environments.

The mechanisms of a wall-climbing walking robot (WCWR) envisage two functions: "attachment" and "locomotion". There are many factors involved in these processes, and one of them is an adhesion method. In this paper, a miniature mobile WCWR with suction cups and an on-board vacuum pump with an additional vacuum tank is discussed. The paper describes a complex simulation model of the suction cup's adhesion to rough surfaces, which can be used for WCWR's conceptual design. The model is based on statistical characteristics of vertical surfaces, statistical characteristics of cracks and on the analysis of interaction between the adhesion system and the surface. The simulation model is used to select the basic parameters of the adhesion system, such as the pump capacity, and the vacuum tank's volume by the Monte-Carlo method. The effectiveness of the robot adhesion system is evaluated with the help of this model. Application of the simulation model is explained on an example. We evaluated the effectiveness of the vacuum tank by the number of robot steps. The results of adhesion system simulation are presented. Besides, we determined the distribution law of the robot's locomotion before it stops. The simulation results reveal that the distribution of the total number of steps follows the gamma-distribution. If we consider only the subsequent steps after the first successful one, their distribution follows the exponential probability law.

This paper presents StructDetect, a fast method for object detection. The target detection process consists of two stages: generation of a hypothesis (object proposals) (1) and verification of the hypothesis (2). Generation of the object proposals is carried out by means of a simple structural model on the basis of line segment combining. Line segment is detected by EdLines algorithm. Then a computer attributes the line segments and their pairs and creates "a connection table", which filters some combinations. Further, it creates a triple combination of the line segments filtered by "the connection table". Each combination has a handcraft descriptor based on the line segment attribute. This descriptor is used to learn kNN classifier and generate object proposals in the area of 3 line segments. These proposals define a set of candidate bounding boxes available to the detector. The second module is based on a convolutional neural network, which takes a fixed-length feature vector from each region. The convolution neural network computes once per image and features vector extracts with adaptively-sized pooling from the last convolution layer. Then the feature vectors are classified by the random forest algorithm. Accuracy of this approach is comparable with the accuracy of such modern detector methods as SPPNet and RCNN. StructDetect is 7 times faster than SPPNet and has a frame rate of 4fps on a CPU.

Sensorless technologies for the field-oriented control of the permanent magnet synchronous motors (PMSMs) are being actively developed today. Recently, a robust, nonlinear and globally convergent position observer for the surface-mount permanent magnet synchronous motor (SPMSM) has been proposed in [4]. The key feature of the robust observer is estimation of the rotor position requiring only the knowledge of the stator currents, voltages, resistance and inductance. However, the resistance and inductance of the stator may change during operation as a result of overheating, mechanical wear and environmental impacts. Thus, the aim of this study is to develop adaptive algorithms for the stator windings' parameters, which can be used for this position observer improvement. In this paper the authors propose four adaptive observers for estimation of the resistance and inductance at low motor speeds. It is assumed, that only currents and voltages of the stator windings are known. The other state variables and parameters of PMSM are unknown. The main idea of the synthesis of identification algorithms is linear filtering, which allows us to eliminate the unknown terms obtained by the mathematical transformations. Requirements to the external load, which ensure successful identification of the PMSM stator parameters, are given for each observer. The proposed adaptive observers for the stator resistance and inductance can be applied in case, when at low speeds, the field-oriented control of PMSM is carried out by means of Hall sensor, while at higher speeds, the sensorless control algorithms are used. A distinctive feature of the proposed estimation algorithms is that the resistance and inductance identification can be implemented during PMSM operation.

The aim of the given research is improvement of the quality of the spark plugs' manufacturing technological process and of their operating properties at different stages of their life cycle. The article is devoted to control of the spark plugs' life cycle through development of a decision support system for the quality of the production process. It demonstrates the correlations between the different stages of the life cycle and spark plugs' manufacturing quality with a view to improve the technological processes, and also identifies the spark plugs' controllable and uncontrollable parameters, affecting the engine unit ignition and starting quality. The article describes an approach to development of a decision support system for the spark plugs' life cycle and certain elements of the system, implemented in real production. The described stages of development of the quality control system for spark plugs' manufacturing technological process and different life cycle stages establish a feedback between the testing and operating steps, and the steps of the product engineering and manufacturing. Within this system certain elements of a subsystem for diagnostics and decision-making were described, and for this purpose an analysis of the faults, defects and reject sources was done. The proposed system structure allows us to solve the problems of the products' parametrical identification at the designing stage and to reveal the problems, which arise during the technological operations. The described approach permits to control the quality of products (sparkplugs) at different stages of the technological process, which makes it possible to reduce the designing and manufacturing costs and to increase the products' quality and competitiveness.

Recently, some articles have appeared which discuss the problem of preflight routing for the unmanned aerial vehicles (UAV). The importance of the problem is determined by the fact that its solving is necessary for flight planning and modeling of UAV with a view of an effective assessment of its mission. The above mentioned papers view routing as a sequence of flights over the setpoints with certain positioning data. The notion of route optimality is associated with, on the one hand, increasing amount of the setpoints included into the route and, on the other hand, with a decrease of the period of time for an en-route flight. In this case, consideration of the problem of routing with a constraint for the en-route flight time is of practical relevance. Thus, it is the amount of the setpoints potentially set by the route that must serve as a maximal criterion. It is quite clear that such statement may result in a situation, when some selected setpoints are out of the optimal route. The paper describes the methods of finding a light UAV circle trip which links the maximum of the equivalent setpoints in a constant wind field taking into account the time constraint. The selected route is the fastest of all the routes linking the setpoints. The methods envisage consecutive solving of the purposely set central and sub-related routing problems. Mathematically, they are the problems of linear Boolean programming. Thus, a flight route can be divided in two stages. At the first stage, it is necessary to define the maximum of the setpoints linked by the routes with the flight time, which does not exceed the one, which is allowed. At the second stage, one should define the fastest of the first stage routes. An example was introduced explaining the details and showing the performance capacity of the present flight routing procedure. The results can be used at the planning stage for modeling of UAV application with a view to improve assessment of its mission effectiveness, as well as for designing of special algorithms and software for the UAV controllers' working area.

Questions related to description and algorithms for device (platform) used to identify space orientation and vertex angle of right circular cone are considered. The main measuring facility - rate gyro attached rigidly to measuring platform is used. Reference frame of platform coincides with gyro frame. For getting solution of the task pointed out above it is suggested to organize three stage process. First-measuring the Earth angular rate; second - carrying the measuring platform by operator from one initial position served as a reference to another desired one (cone surface) and third-performing measurements by relocating the platform on cone surface without loss of contact. The most important stage - performing the measurements. To try out the third stage it was worked out an environmental simple software-based simulator. To delete disturbances from operator activity it is suggested to measure the vertical axis angular direction of reference platform frame in current platform position. So properties of dot and cross products for platform different angular positions (for vertical axis only) are used together with procedure of kinematic equation integration. Mathematical description and processing algorithms are given to perform on-line treatment of current measurement data and get the estimates of orientation and vertex angle. Experiments were performed on software-based simulator which showed good results. It is turned out the suggested approach may be extended to more wider area of application as estimation of mutual orthogonality of separate construction elements and so on.