The detection system of the obstacles on the road, data fusion multi-sensor (emergency braking system of the car)

Dissertation for the degree of M.Sc. Mechatronics.

Abstract:

The subject of this research is the design and simulation of data integration in a radar network that has overlapping. Data fusion means combining the output data of dissimilar radar sensors that differ in terms of accuracy in ranging and angle measurement. These sensors are installed in a row in front of the car to better detect obstacles on the road. Each radar sends the data related to its report of the target position to the data integration center, by applying the data integration algorithm to this data, a more accurate estimate of the location and speed of the target can be achieved. The method of doing the work is that by applying the Kalman filter estimation method and the developed Kalman filter on the information sent from the sensors, the amount of error has been reduced, and then all kinds of sensor information integration methods (hierarchical and collective) have been examined at two levels of observation data integration and tracking data integration. The purpose of this research is to integrate the tracking data in order to integrate the coverage of the area and improve the accuracy of the target position estimation after the tracking stage. We have used the mean square error criterion to evaluate the system. The results showed that the positioning error of each of the sensors is dependent on the target position relative to the radar in addition to the accuracy of the radar. Moreover, with the increase in the number of sensors, the positioning accuracy has increased and gradually the accuracy of the target positioning has improved with the increase in the number of sensors, so that the positioning error of the integration center at any moment is better than the positioning error of individual sensors that overlap at that moment.

Key words:

Data fusion - Vehicle emergency braking - Kalman filter - Extended Kalman filter - Radar target tracking - Situational awareness - Combination of sensors

Introduction:

Benefits of combination Data from multiple sources over single-source data has created a research process for several decades in the field of multi-source and multi-purpose data integration process. Hall and James have presented a comprehensive and complete introduction about this topic, its application areas, architectural model and its implementation methods. The final goal of the data integration process is to increase the accuracy in the process of understanding the observed phenomenon as well as inferring future scenarios.]5[,]6[

What is information integration?

We will start with a simple example, consider the organs and parts of a human, assume each part as a sensor that sends information to the human brain, is processed in the brain and applies the appropriate output. slow down The aspect of humans is input from the five sense organs: touch, smell, taste (tasting), hearing, vision and different physical structures by an incredible process, which is not yet fully understood, humans transfer the input from these organs through the brain to feeling in reality. We need to feel or be sure that we are in a place, in coordinates, in place, in time. Thus, we have obtained a more complete picture of an observed sensation. The human activities of design, art, investment, market analysis, military intelligence, complex artwork, complex dance sequences, music creation, and journalism are good examples of activities that use the concepts and aspects of advanced DF data synthesis that we do not yet fully understand. Perhaps, the human brain integrates such data or information without any automatic assistance, because it has an associative reasoning ability developed over thousands of years. The work of data fusion is the work of the brain that combines information and deduces new information from it that could not be deduced by any sensor before.

As a second example, to better understand the integration of information, suppose we have two sensors, one is a radar that determines where the enemy plane is located, but does not give us any information in terms of altitude (some radars have such a capability, here just as an example and for a better understanding, it is assumed that we do not have such a radar) and an infrared sensor that gives us the altitude, but does not give us any information about the location of the enemy. Combine these two and the main advantage of data fusion becomes apparent. In this case, we can infer and extract information about the location and position of the enemy aircraft and many other information that is extracted from the total information of these two sensors, such as its speed, arrangement and behavior, etc.

What is clear is that when information is combined, more information is produced, and more information means more power.

Data fusion has many uses, including the following. :

Transportation, air navigation (military and passenger), manufacturing of smart cars, road traffic management, multimedia telecommunications, combination of audio and video in remote mass communication, smart home systems, robotics, 3D displays, calculation of material corrosion, etc.

One ??of the terms that we come across a lot is the term fuse, which according to the example above is what the brain does when It includes receiving several signals from sensors, for example, processing them and producing the appropriate output. Now if we extend the same meaning to data fusion, it won't be too difficult.

1-1) What is the data fusion processing framework?

A framework that tells us how and where to start, what to do and what to do to work with sensors that are serving us centrally. Let's get some results. For example, in software engineering, we have many methodologies by which we manage the design and development of our software. In the present discussion, the frameworks are the same methodologies in software engineering. One of the most famous frameworks is the JDL framework [1], which literally means the combination and interconnection of laboratory managers within the United States Defense Organization. It is designed for military work. becomes:

Level one: defines objects. For this definition, a general picture of the existing position of that object should be reported. This work is done by combining the features and characteristics of that object that are extracted from several sources or sensors or any other source. We state this definition more clearly from another source: this stage is the task of mixing and linking different spatial, temporal and... information in order to obtain an object that is refined and unique, such as a weapon, a location or a geographical military unit or anything else as a neat object.  It does this by relating other entities that are important to our collection as well as observed events. Another definition of [6] is stated: This stage defines a description and explanation of our connections between objects and events in the heart of their surrounding environment. Objects that are distributed and obtained from the first part or level one are examined to become a meaningful unit. In general, this stage deals with communication information between objects. There are various types of communication, such as physical communication, the purpose of which is to determine the meaning and concept of a set of entities. This analysis is in various fields such as land, surrounding media, hydrology, weather and other factors. This stage calculates the advantages and disadvantages of one period of operation compared to another period (for example, the previous period). In other words, in another source, this stage is responsible for planning the current situation for the future. In fact, it argues and infers what the target's behavior or its vulnerabilities will be like, and it argues these behaviors [1], [7].