Search

The creation of an unmanned vehicle is an extremely complex task in which it is necessary to think through a lot of technical aspects and levels of abstraction, from working with the hardware (motors, motion controllers, cameras, motion sensors, body), ending with working with AI (developing neural networks, choosing the appropriate models, training, tuning). All this takes up a lot of development resources, both material and temporary. The purpose of this graduate work is to create a unique traffic control and forecasting module specifically for the developers of the field of unmanned vehicles - to create a certain platform that can several times reduce the number of resources spent on developing prototypes of unmanned vehicles.

The idea of replacing a living person with artificial intelligence is far from new. One of the many areas where this idea can become a reality is in self-driving vehicles. The creation of an unmanned vehicle is an extremely complex task in which it is necessary to think through a lot of technical aspects and levels of abstraction, from working with the hardware (motors, motion controllers, cameras, motion sensors, body), ending with working with AI (developing neural networks, choosing the appropriate models, training, tuning). All this takes up a lot of development resources, both material and temporary. The purpose of this work is to create a unique module for traffic control and forecasting specifically for developers of the field of transport - to create a certain platform that can several times reduce the amount of resources spent on the development of prototypes of unmanned vehicles.

The technique of calculation of vehicle weight with the use of discrete piezoelectric sensor for automatic weight and dimensional control systems of vehicles, allowing to make calculations with the least errors in comparison with the existing monocrystal analogues, is presented. Improving the accuracy of measurements is necessary to improve the quality of the procedure for assessing the physical characteristics of the vehicle at the points of automatic weight and dimensional control

The article presents the current problems related to optimization of MSW (municipal solid waste) transportation and sorting. The relevance of this topic in the modern world is at the highest level, the amount of garbage is growing, and territories are limited, which leads to the issue of waste recycling. One of the modern technologies for automation of waste recycling plants was considered. And also, the economic aspect of processing, with the possibility of excluding a person from the process of monitoring the performance of the technological process.

This work is devoted to the study of crack detection algorithms in the roadway, and the development of a system of automatic defect detection using computer vision and neural network. The article provides a classification of defects, as well as a classification of the possible causes of their formation. A review of the work related to this problem area is given. Two alternative solutions are proposed, based on a convolutional neural network and a feed forward network. These solutions are used as a first step of crack classification.

The article is devoted to the development of a client-server application for detecting road infrastructure objects. The main architectures of neural networks are considered, a database for collecting the necessary data is developed. A feature of the developed application is the binding of road infrastructure objects with geo-position and the formation of a map with the display of recognizable objects.

This article presented the element base and an electronic filter with an amplifier, sbRIO-9631 programmable logic integrated circuit, which allows you to read the signal from a sensor, amplified and filtered. In turn, the use of FPGAs will allow you to flexibly and quickly expand the system of future needs. The possibility of further modification is also described, due to the large number of input and output ports and the FPGA itself.

Currently, the theory of pattern recognition is used in many areas of human life. The use of such technologies can improve the efficiency of solving problems and helps reduce the number of errors based on the human factor. In this paper, the basic methods of object recognition are considered, the creation of a software product for solving the problem of detecting and recognizing images is described. The developed application is multithreaded, which allows efficient processing of large volumes of information. The application is designed for operating systems of the Windows family, based on the. NET Framework.

In this paper, the idea of processing the object of observation using stereo vision systems is advanced. Stereoscopic vision is the direction of machine vision that allows you to create a three-dimensional picture of the world around you. In laboratories around the world, practical research is being conducted on the use of stereo vision for solving applied problems. Stereoscopic vision repeats the capabilities of the human eye, allowing you to get not only color and brightness from the input image, but also the distance to it. This technology will allow you to do without expensive equipment, sensors, radars for monitoring traffic and fixing violations.

This article is devoted to modern information and communication methods of weight and dimensional control over time. The paper analyzes the existing systems, discusses the problems of determining the mass in dynamics and methods for solving them using signal filtering.

The latest direction in a research of construction composite materials having the term "computer materials science" which is engaged in studying of properties of composite materials on the basis of modeling with use of computers is presented in article. The results of this work allow you to simulate materials for absorbing different types of waves that can be used in the construction of shared roads.

The article discusses the influence of the type of suspension of the universal spreader of the de-icing material installed on the mobile chassis of the combined road machine on the dosing accuracy. Since the change in environmental parameters is continuous, automated road meteorological stations are used for greater efficiency, and on the basis of these data, it is possible to quickly change the process of distribution of the deicing material by combined road machines.

The purpose of the work is to study the typification methods of moving objects based on a neural network. It turned out that the use of image classifications is so widespread today that it is difficult to imagine a sphere in which any algorithms are not implemented. The task of the work is developing system for detecting road signs on the video.

This article has a system of automatic weight and dimensional control (weight - and - motion), which allows weighing, measuring geometric characteristics, as well as recognition of state registration plates, passing vehicles without stopping, as well as processing of this data and registered violators. The focus is on the work of measuring the mass along the axes. All conversions are performed on the National Instruments multifunction board, which has a microcontroller, additional I/O peripherals field-programmable gate array, which means user-programmable gate arrays. In a more general case, they are called FPGAs - programmable logic integrated circuits. This study is important because heavy vehicles smash federal and local roads. Therefore, it is necessary to control the passage, however, now this measurement is carried out only in static mode. Measurement of dimensions in dynamics based on piezoelectric elements has an error unacceptable by the legislation of the Russian Federation. The developed method will increase the accuracy, which will lead to a mass measurement of dimensions, and will allow to maintain in good condition the roads of federal and local destination.