KR102266184B1 - Control device for piloting aircraft and operation method thereof - Google Patents

Abstract

The present disclosure relates to a control device for maneuvering an aircraft and a method of operating the same. According to the present disclosure, the operation method may comprise the steps of: obtaining at least one of data about the aircraft and data about the environment around the aircraft; extracting a state value for action determination from the obtained data; determining the maneuvering behavior of the aircraft from the state value based on a behavior model based on the fuzzy rule and the behavior tree structure; generating a tracking point corresponding to the maneuvering behavior; and controlling the maneuvering of the aircraft according to the tracking point.

Description

CONTROL DEVICE FOR PILOTING AIRCRAFT AND OPERATION METHOD THEREOF

The present disclosure relates to a control device for maneuvering an aircraft and an operating method thereof.

The need to develop artificial intelligence control technology is increasing day by day. If the existing manned squadron is improved into a manned and unmanned squadron form, each trained pilot can not only control their own aircraft, but can control a single manned and unmanned squadron mixed with AI pilots, which means that the same number of trained pilots can be operated. As pilots can control a much larger number of aircraft, air power can be expected to increase through efficient formation of squadrons. To meet this need for development, the US Defense Advanced Research Institute (DARPA) is promoting the development of human-level artificial intelligence for reliable and scalable air combat through the ACE (Air Combat Evolution) program. Alpha Dogfight Trials are being conducted for technology exchange and efficient development.

Control for artificial intelligence operation of aircraft can be divided into three parts: Decision, Guidance, and Control. Judgment determines which maneuver is advantageous based on the current state to determine the tracking point. Induction generates a steering value for moving the aircraft to the corresponding tracking point, and the control derives the next state by simulating the aircraft using the steering value.

Looking at the related technologies, first, there are technologies that directly perform the movement of the aircraft from the current state without intermediate processes such as generation of tracking points and control values, and methods such as Minmax tree search, Reinforcement Learning (RL), and genetic Fuzzy trees and heuristic techniques were used. In addition, there is a technology that directly uses fuzzy rules to determine what kind of thrust a fighter should have from an immediate state without the concept of a tracking point. However, this technology has not been verified as to which control input value should be used to obtain the output thrust, and how the actual fighter maneuver will occur depending on the maneuvering dynamics or the physics engine when it is input. In terms of the complexity of the overall system in common, the aforementioned technologies are simple compared to the structures divided into judgment, guidance, and control, but in the end, what kind of input must be taken from the pilot's point of view at the final control stage to move the aircraft along the selected path? is not given, and verification of actual maneuvers based on dynamics or physical models has not been carried out, so it is somewhat difficult to apply them immediately. In addition, in the case of Minmax tree search, the expected value of the final result when each option given in the current situation is selected numerically is calculated, and then the selection is made in the direction of minimizing the win rate of the bandit and maximizing the win rate of the ally. There is a disadvantage that the amount of calculation increases as the numbers up to a few numbers are calculated one by one, so there is a limit to applying it to an air combat situation where a decision must be made quickly at every moment.

Contrary to the above, there are also related technologies that classify the control stage and verify the operation in the actual situation/physical model. There is also a technique that binds the judgment and guidance parts together and outputs what kind of thrust the ally should receive from a given state, and uses this to verify the actual operation. Here, Minmax tree search is used in the front part, and MATLAB in the control part in the back part fmincon function-based optimization of Although it has improved compared to the above-mentioned techniques in that it adds calculation in the control part, the problem due to the high amount of computation still remains because the Minmax tree search is used in the previous part.

Disclosed embodiments are intended to disclose a control device for maneuvering an aircraft and an operating method thereof. Specifically, the embodiment according to the present disclosure divides the overall structure for controlling the movement of the fighter from the state value into three parts, largely judging, guiding, and controlling, and designing the judgment part based on the manual to solve the above problem It can be designed as a behavior model that applies fuzzy rules to the behavior tree structure and decision process of the behavior tree and optimizes detailed parameters through machine learning. The technical problem to be achieved by the present embodiment is not limited to the technical problems as described above, and other technical problems may be inferred from the following embodiments.

In order to achieve the above object, a method for controlling the operation of an aircraft according to an embodiment of the present specification includes: acquiring at least one of data about the aircraft and data about the environment around the aircraft; extracting a state value for action determination from the obtained data; determining the maneuvering behavior of the aircraft from the state value based on a behavior model based on a fuzzy rule and a behavior tree structure; generating a tracking point corresponding to the maneuvering action; and controlling the operation of the aircraft according to the tracking point.

According to an embodiment, the behavior model may be based on a behavior tree structure modeled based on a manual, and at least one determination process may be implemented in the behavior model according to the fuzzy rule.

According to an embodiment, at least one of a member function and a fuzzy variable related to the fuzzy rule may be optimized through machine learning.

According to an embodiment, the machine learning may include a genetic algorithm.

According to an embodiment, optimizing the fuzzy variable may include changing a fuzzy variable affecting the maneuvering behavior.

According to an embodiment, the controlling of the operation of the aircraft may include: generating a control command value of the aircraft from the tracking point; and performing a piloting simulation of the aircraft based on the steering command value.

According to an embodiment, the state value may be updated based on a result of the steering simulation.

According to an embodiment, the state value may include at least one of a distance between a plurality of aircraft, a side angle, a nose crossing angle, and an antenna aiming angle.

According to an embodiment, the determining of the maneuvering behavior of the aircraft may include determining the maneuvering behavior for each preset period, and the period may be set differently according to a change amount of the acquired data.

In order to achieve the above object, a non-transitory computer-readable storage medium according to an embodiment of the present specification includes a medium configured to store computer-readable instructions, wherein the computer-readable instructions are executed by a processor. obtaining, by the processor, at least one of data about the aircraft and data about the environment around the aircraft; extracting a state value for action determination from the obtained data; determining the maneuvering behavior of the aircraft from the state value based on a behavior model based on a fuzzy rule and a behavior tree structure; generating a tracking point corresponding to the maneuvering action; and controlling the operation of the aircraft according to the tracking point.

In order to achieve the above object, an apparatus for controlling a flight of an aircraft according to an embodiment of the present specification includes: a memory for storing at least one instruction; and by executing the at least one command, obtain at least one of data about the aircraft and data about the environment around the aircraft, extract a state value for action determination from the obtained data, and have a fuzzy rule and action tree structure A processor that determines the maneuvering behavior of the aircraft based on the behavior model based on the state value, generates a tracking point corresponding to the maneuvering behavior, and controls the operation of the aircraft according to the tracking point. may include

According to the embodiment of the present specification, by adopting a structure in which fuzzy rules and machine learning are applied to the action tree in the decision process for determining the maneuver of the fighter, the pilot at the time of complex operation with or without explainable artificial intelligence that defines the rules based on the manual and operators can be trusted. In addition, by applying the genetic fuzzy rule that combines the fuzzy rule and the optimization through genetic algorithm to the ambiguous part and the part that needs optimization in the existing manual, the performance of the AI control system can be improved, and depending on the characteristics of the manual-based system, the performance of the AI control system can be improved. It can facilitate the expansion of the mission according to the addition of rules. In addition, it is possible to secure guidance stability and steering stability by separating the judgment and control parts.

Effects of the embodiment are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the description of the claims.

1 shows a control device according to the present disclosure.
2 is a diagram illustrating positional geometry information between a first aircraft and a second aircraft according to the present disclosure.
3 is a diagram illustrating a schematic process of designing a behavior model according to the present disclosure.
4 shows a structure of an action tree designed by analyzing the manual according to the present disclosure.
5 is an exemplary diagram for describing a characteristic of a fuzzy rule according to the present disclosure.
6A to 6C are exemplary diagrams for explaining a process of optimizing at least one of a member function and a fuzzy variable related to a fuzzy rule according to the present disclosure through machine learning.
7 is an exemplary diagram illustrating a process for optimizing a member function for a fuzzy rule using a genetic algorithm according to the present disclosure.
8 is a diagram schematically illustrating a process in which a steering control model and a flight control model according to the present disclosure operate.
9 is an exemplary diagram illustrating tracking point guidance control of an aircraft according to the present disclosure.
10 shows an overall architecture representing a process by which an apparatus for controlling the steering of an aircraft according to the present disclosure operates.
11 illustrates a method of operating a control device according to the present disclosure.

Terms used in the embodiments are selected as currently widely used general terms as possible while considering functions in the present disclosure, but may vary according to intentions or precedents of those of ordinary skill in the art, emergence of new technologies, and the like. In addition, in a specific case, there is a term arbitrarily selected by the applicant, and in this case, the meaning will be described in detail in the corresponding description. Therefore, the terms used in the present disclosure should be defined based on the meaning of the term and the contents of the present disclosure, rather than the simple name of the term.

In the entire specification, when a part “includes” a certain component, it means that other components may be further included, rather than excluding other components, unless otherwise stated. In addition, terms such as “~ unit” and “~ module” described in the specification mean a unit that processes at least one function or operation, which may be implemented as hardware or software, or a combination of hardware and software.

The expression “at least one of a, b, and c” described throughout the specification is, 'a alone', 'b alone', 'c alone', 'a and b', 'a and c', 'b and c ', or 'all of a, b, and c'.

Hereinafter, with reference to the accompanying drawings, embodiments of the present disclosure will be described in detail so that those of ordinary skill in the art to which the present disclosure pertains can easily implement them. However, the present disclosure may be implemented in several different forms and is not limited to the embodiments described herein.

Hereinafter, embodiments of the present disclosure will be described in detail with reference to the drawings.

1 shows a control device according to the present disclosure.

The control device 100 includes a processor 110 and a memory 120 . In the control device 100 illustrated in FIG. 1 , only components related to the present exemplary embodiments are illustrated. Accordingly, it is apparent to those skilled in the art that the control device 100 may further include other general-purpose components in addition to the components illustrated in FIG. 1 .

The control device 100 may control the operation of the aircraft. Specifically, the control device 100 determines an appropriate maneuvering action by recognizing the surrounding situation, generates a tracking point for performing the determined starting action, and receives a steering command value (stick, throttle, rudder) for moving to the tracking point. can decide

The processor 110 serves to control overall functions for the operation of the aircraft in the control device 100 . For example, the processor 110 generally controls the control device 100 by executing programs stored in the memory 120 in the control device 100 . The processor 110 may be implemented as a central processing unit (CPU), a graphics processing unit (GPU), an application processor (AP), etc. provided in the control device 100 , but is not limited thereto.

The memory 120 is hardware for storing various types of data processed in the control device 100 , and the memory 120 may store data processed by the control device 100 and data to be processed. Also, the memory 120 may store applications, drivers, and the like to be driven by the control device 100 . The memory 120 includes random access memory (RAM), such as dynamic random access memory (DRAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), CD- It may include ROM, Blu-ray or other optical disk storage, a hard disk drive (HDD), a solid state drive (SSD), or flash memory.

The processor 110 may acquire at least one of data about the aircraft and data about the environment around the aircraft. In an embodiment, the processor 110 may obtain data about the aircraft (eg, the position of the aircraft, the amount of fuel) and data about the surrounding environment (eg, the position of the bandit) through the sensor. The sensor may include a vision sensor, a RADAR sensor, a GPS/INS sensor, and the like, and the type and quality of data that can be acquired may vary depending on the type of sensor. The acquired data may be stored in the memory 120 .

The processor 110 may extract a state value for action determination from the obtained data. In an embodiment, the state value for determining the behavior may be feature data extracted from acquired sensor data, and the feature data includes positional geometry information indicating the relative position between the aircraft and other aircraft, remaining fuel energy, current speed data, etc. may include In an embodiment, the processor 110 may identify the position of another aircraft from image data obtained from a sensor through an image recognition technique and extract position geometry information as a state value. A state value for determining an action may be stored in the memory 120 .

2 is a diagram illustrating positional geometry information between a first aircraft and a second aircraft according to the present disclosure.

Referring to FIG. 2 , the first aircraft may be a friendly aircraft and the second aircraft may be a bandit aircraft. The location geometry information between the first aircraft and the second aircraft may include a distance (Range), an aspect angle (AA), an antenna train angle (ATA), and a head crossing angle (HCA). can The distance (Range) is a distance between the first aircraft and the second aircraft, and may correspond to the length of a virtual line connecting the first aircraft and the second aircraft. The side angle may correspond to an angle between an imaginary line connecting the first aircraft and the second aircraft and a traveling direction of the second aircraft. The antenna aiming angle may correspond to an angle between an imaginary line connecting the first aircraft and the second aircraft and a traveling direction of the first aircraft. The nose crossing angle may correspond to an angle between the traveling direction of the first aircraft and the traveling direction of the second aircraft. Such location geometry information may be extracted from the sensor data obtained by the processor 110 and input to a behavior model to be described later in order to determine a maneuvering behavior suitable for a situation.

Referring back to FIG. 1 , according to an embodiment, the processor 110 may determine the maneuvering behavior of the aircraft from the state value based on a behavior model based on a fuzzy rule and a behavior tree structure. have.

3 is a diagram illustrating a schematic process of designing a behavior model according to the present disclosure.

The overall structure of the system for controlling the movement of the fighter from the state value can be divided into three parts: judgment, guidance, and control. In an embodiment, the determination part may be implemented using an artificial neural network (ANN). In one embodiment, the decision part may be implemented using a Naive Bayesian model. In one embodiment, the derivation portion may be implemented using a Pursuit Point Guidance algorithm. In an embodiment, the decision part may be implemented based on a behavior model based on a behavior tree structure and a fuzzy rule, as will be described later.

The behavior model can be used to output the optimal maneuvering behavior to favor the battle based on the state value for the behavior judgment obtained earlier. Behavior models can be designed using behavior tree structures and fuzzy rules. The behavior model is based on a behavior tree structure that is modeled based on a manual, and at least one decision process in the behavior model may be implemented according to the fuzzy rule. The action tree means a tree structure designed for action judgment based on the analysis of the manual, and the fuzzy rule expresses the parts vaguely described in the manual (for example, judgment of superiority in the state of the bandit) It is a logical structure that can be used to make decisions under conditions vaguely described in the manual in an action tree structure. The initial model designed using the behavior tree structure and fuzzy rules can be optimized through machine learning, and machine learning is a representative technique of Evolutional Computation that mimics the evolution of living things. can

Hereinafter, an action tree and a fuzzy rule for designing a behavior model will be described in detail.

4 shows a structure of an action tree designed by analyzing the manual according to the present disclosure.

Referring to FIG. 4 , an aircraft pilot training manual 410 and an action tree structure 420 modeled by analyzing the manual 410 according to an embodiment are shown.

The manual 410 may be used when actual aircraft pilots receive flight training. In one embodiment, the manual 410 may include content for Basic Fighter Maneuvers (BFM) that describes the tactical maneuvers that a fighter performs to gain an advantage over the bandit during an aerial engagement. In one embodiment, the manual 410 may describe a method of determining a maneuver by recognizing the superiority, inferiority, and equivalence situation of the friendly plane from the positional geometry information between the friendly plane and the bandit. The information used to determine the situation may include position geometry information such as the current position of the friendly flag and the bandit and the relative distance between the friendly flag and the bandit shown in FIG. 2 , the side angle, the nose crossing angle, and the antenna aiming angle. In one embodiment, the manual 410 provides the content of the BFM for one-to-one engagements of fighters as well as Air Combat Maneuvers (ACM) for many-to-one or many-to-many engagements of fighters. ) may be included.

The action tree structure 420 may be a structure modeled as an action tree by analyzing the manual 410 and the knowledge of the professional pilot, and based on the analysis, which maneuver the pilot should decide for each situation with the bandit in the engagement situation. . The behavior tree structure 420 is a structure used when implementing artificial intelligence in computer engineering, robotics, control systems, game systems, and the like. The action tree structure 420 may include a root node 421 that is a topmost node, a control flow node 422 that determines a condition, and an execution node 423 that executes the action. The processor 110 starts at the root node 421 , and determines whether state values such as position geometry information, remaining fuel energy, and turning radius satisfy specific conditions in the control flow node 422 , and a lower execution node 423 . may determine whether to execute, and the execution node 423 may include maneuvers performed by the aircraft, such as lead tracking, lag tracking, and pure tracking. For example, the friendly aircraft determines whether it is advantageous/equal/disadvantaged in the current relationship with the bandit through the positional relationship with the bandit identified through situational awareness, and whether it is currently attacking/defending/neutral, and in each case Select OBFM (Basic Offensive Combat Maneuver), DBFM (Defense Basic Combat Maneuver), and HABFM (Front Basic Combat Maneuver), and detailed maneuvers can be selected at the lower level of the action tree corresponding to each BFM.

For example, if the manual 410 states, "Inside the turning radius of the target, turning to a smaller radius than the target means more energy consumption than the target." It is determined whether the flow node 422 satisfies the condition of “small turning radius AND low energy”, and if it is satisfied, it is determined to execute a “lag-tracking” maneuver in the execution node 423 of the lower level, so that the tracking point behind the defender can be modeled to create In another example, the manual (410) states, "Even if the turning radius ability of the attacker is slightly lower than that of the defender, the turn rate using energy can continue to turn on a par with the defender who can turn with a smaller radius." , the action tree structure 420 determines whether the condition of “turn radius similar AND energy high” is satisfied in the control flow node 422, and if this is satisfied, the “lead” in the lower-level execution node 423 It can be modeled to create a tracking point in front of the defender by deciding to execute a “track” maneuver.

In an embodiment, in determining the maneuvering behavior of the aircraft through the behavior model, the processor 110 may determine the maneuvering behavior for each preset period, and the period may be set differently according to the amount of change in the acquired data. If the processor 110 recognizes the surrounding situation and sets the cycle for determining the maneuvering action accordingly, a more suitable maneuver can be determined in a battle situation requiring a quick decision, and the risk of an accident that may occur is reduced, but it is inefficient due to the large amount of calculations. can Therefore, if the amount of change in the acquired data is small, it is likely that the aircraft is moving slowly or there is little change in the surrounding situation. It is possible to reduce the overload of the processor 110 and promote efficiency.

5 is an exemplary diagram for describing a characteristic of a fuzzy rule according to the present disclosure.

Referring to FIG. 5 , characteristics of a general Boolean rule and a fuzzy rule are contrasted in describing information. Fuzzy rule, or fuzzy logic, is a logical structure that expresses an ambiguous state or an ambiguous state as multivalued, away from the binary logic of true or false. For example, if the aircraft has about 60% of fuel remaining, the Boolean rule would simply state that the remaining fuel is "moderate" or "enough" fuel, whereas the fuzzy rule, for example, "fuel low" 0, "Fuel Moderate" can be expressed as 0.6, and "Fuel Sufficient" as 0.4. A fuzzy rule may be composed of a fuzzy variable representing information to be expressed and a membership function defining a value of the fuzzy variable.

In the example 510 of FIG. 5 , the turn radius R may be calculated as a ratio relative to the bandit's turn radius. In the Boolean rule, _{if the turn radius R is less than r 1} (e.g. 80% of the bandit), it is expressed as "the turn radius R is small", and the turn radius R is greater than or equal _{to r 1 and equal to r 2} (e.g., If it is less than 120% of the bandit), _{it can be expressed as "the turning radius R is similar", and when the turning radius R is larger than r 2} , it can be expressed as "the turning radius R is large". In the fuzzy rule, the turning radius R is set as a fuzzy variable, and various fuzzy sets such as "small", "similar", and "large" of the turning radius can be defined for the same fuzzy variable, and each fuzzy A set may be defined as a member function as shown in graph 511 . The member functions defining each fuzzy set may have overlapping sections as shown in the graph 511. For example, when the turning radius is about 80% compared to the bandit, the turning radius is expressed as small compared to the bandit. The turn radius may be expressed as similar to the bandit, and the degree of turning radius smaller than the bandit may be expressed as 0.7, and the degree similar to the bandit may be expressed as 0.4.

Likewise, in another example 520 of FIG. 5 , the energy of an aircraft may also be expressed in terms of a Boolean rule as well as a fuzzy rule. In the fuzzy rule, energy E is set as a fuzzy variable, and various fuzzy sets such as energy “high”, “normal”, and “low” can be defined for the same fuzzy variable, and each fuzzy set is shown in the graph 521 . As shown, it can be defined as a member function.

In this way, the fuzzy rule breaks away from the binary logic that each object belongs to or does not belong to any set, and the degree to which each object belongs to the set can be expressed as a member function, so the manual 410 is modeled as an action tree structure 420. In this case, the parts vaguely described on the manual 410 (eg, advantageous judgment, the defender tracking status of the bandit (attacker) (Lead, Lag, Pure) judgment, the defender's It has the advantage of being able to adaptively express whether a break turn succeeds or not, and whether an attacker sets a virtual tracking path, etc.).

Hereinafter, a method for optimizing a behavior model designed using the behavior tree structure and fuzzy rules through machine learning will be described in detail.

In an embodiment, in a behavior model based on a fuzzy rule and a behavior tree structure, at least one of a member function and a fuzzy variable related to a fuzzy rule may be optimized through machine learning. In a behavior model in which the judgment process in the behavior tree is implemented according to a fuzzy rule, which fuzzy variable to use as a judgment factor and how to determine the member function that defines each variable will have a significant impact on the performance of the fuzzy rule-based system. can In the present embodiment, the processor 110 is to be able to find the optimal maneuver that can advantageously fight the enemy in a given situation by the behavior model through machine learning for member functions or fuzzy variables related to fuzzy rules. can For example, machine learning tunes fuzzy variables and member functions on the fuzzy rules of a behavior model to obtain better results by training the behavior model with a training data set containing multiple state values input to the behavior model. can do.

6A to 6C are exemplary diagrams for explaining a process of optimizing at least one of a member function and a fuzzy variable related to a fuzzy rule according to the present disclosure through machine learning.

6A is an exemplary diagram illustrating an initial fuzzy variable set as a variable capable of determining the maneuvering of an aircraft, and an initial member function for a fuzzy set of each fuzzy variable. In the example shown in Fig. 6a, the turn radius and energy of the aircraft were set as initial fuzzy variables, the fuzzy sets of turning radii "small", "similar", "larger", and the fuzzy sets of energy "low", The initial member functions for "Normal" and "High" are shown respectively.

6B shows an exemplary result in which a member function defining a fuzzy variable is optimized through machine learning. In the example shown in FIG. 6B , member functions defining each fuzzy set for fuzzy variables such as turning radius and energy were changed through machine learning. The section value of the fuzzy set defining each fuzzy set for the fuzzy variables, turning radius and energy, and the shape of the graph were changed.

6C shows another exemplary result in which a fuzzy variable and a member function defining a fuzzy variable are optimized through machine learning. In the example shown in FIG. 6C, member functions defining each fuzzy set for fuzzy variables such as turning radius and energy were changed through machine learning. The section value of the fuzzy set defining each fuzzy set for the fuzzy variables, turning radius and energy, and the shape of the graph were changed. Also, in one embodiment, as shown in FIG. 6C , optimizing the fuzzy variable may include changing the fuzzy variable affecting the maneuvering behavior. In this example, the distance between the aircraft is added through machine learning as a fuzzy variable that can determine the maneuvering of the aircraft. As such, the types of variables that affect judgment in a specific control flow node of the behavior tree can also be added or removed by optimizing them through machine learning.

Although the process of optimizing at least one of a member function or a fuzzy variable related to a fuzzy rule through machine learning has been described with reference to FIGS. 6A to 6C , it is not intended to limit the configuration of the present invention to those skilled in the art. will understand

In one embodiment, machine learning for optimizing at least one of a fuzzy variable or a member function for a fuzzy rule may include a genetic algorithm. A genetic algorithm is a computational model based on the evolutionary process of the natural world. It is a representative technique of evolutionary computation that mimics the evolution of living things. Global optimization employs many parts in the evolutionary process of the real natural world, such as selection, crossover, mutation, and substitution. is a technique When solving a problem through a genetic algorithm, the solution can be expressed in the form of a gene or chromosome, and how appropriate this solution is can be calculated through a fitness function.

7 is an exemplary diagram illustrating a process for optimizing a member function for a fuzzy rule using a genetic algorithm according to the present disclosure.

Referring to FIG. 7 , a process of applying a fuzzy rule and a genetic algorithm in determining an optimal maneuver by determining whether a friendly aircraft is advantageous, disadvantageous, or equal according to the relative energy and relative altitude of the aircraft according to an embodiment of the present invention is shown

In this embodiment, "It is advantageous if the relative energy is large (L) and the relative altitude is large (L)", "If the relative energy is moderate (M) and the relative altitude is large (L), it is equal", "Relative In order to make a judgment such as "It is disadvantageous if the energy is small (S) and the relative altitude is large (L)," the relative energy and the relative altitude are set as fuzzy variables for judging the advantage, equal, and disadvantage of the friendly plane first, and the fuzzy Member functions can be defined for the variables relative energy and relative altitude. When defining a member function, the interval value of the member function defining each fuzzy set ("large(L)", "medium(M)", "small(S)") for the relative energies and relative elevations of the fuzzy variables. An initial population of (X ₁ , X ₂ , X ₃ ) can be established, and a plurality of chromosomes (Chromosome 1, Chromosome 2, ... , Choromosome N) with different interval values can be generated. The processor 110 may select an excellent chromosome by applying a fitness function to the initially generated N chromosomes, and may configure a new generation by applying crossover and mutation between excellent chromosomes. The processor 110 may select an optimal chromosome by repeating the evaluation through such a fitness function over generations. Based on the interval value of the member function of the relative energy and the relative altitude corresponding to the optimal chromosome, it is determined whether the ally is advantageous, unfavorable, or equal from the state value of the relative energy and the relative altitude, and the maneuvering behavior is determined accordingly can For example, if the friendly flag is judged to be advantageous or equal, a tracking point located 3/9 behind the enemy can be created with Two Circle, and a tracking point located 3/9 behind the enemy can be created with One Circle when it is judged that the friendly plane is unfavorable. can create

Referring back to FIG. 1 , according to an embodiment, the processor 110 may generate a tracking point corresponding to the maneuvering behavior determined in the behavior model, and control the operation of the aircraft according to the generated tracking point. The processor 110 may generate a tracking point for performing a maneuvering action previously determined based on the behavior model through the steering control model, and may derive a steering command value for controlling the operation of the aircraft according to this.

8 is a diagram schematically illustrating a process in which a steering control model and a flight control model according to the present disclosure operate.

Referring to FIG. 8 , the processor 110 may receive a target maneuver to perform a maneuver selected from the behavior model through the steering control model 810 , generate a tracking point, and finally generate a steering command value. As described above, in the behavior model, advantageous tactical/mobile actions are determined according to the situational judgment of the friendly fighter's bandit, and in the control control model, it indicates to which position the fighter should actually move to take the input tactical/maneuvering action. It can be set as a tracking point. In the generation of the tracking point, not only the maneuvering behavior determined from the behavior model, but also a state value for behavior determination including position geometry between a plurality of fighters obtained through actual flight or simulator implementation may be used as input. In addition, the energy management of the aircraft, the turning radius and its center point, and the weapon behavior characteristics may also be considered in the generation of the tracking point. The steering control model 810 derives values such as a stick, a throttle, and a rudder as a steering command value for moving along the generated tracking point, and provides it to the flight control model 820. have.

9 is an exemplary diagram illustrating tracking point guidance control of an aircraft according to the present disclosure.

Referring to FIG. 9 and an enlarged view 900 thereof, a tracking point indicating a position where the aircraft needs to move in order to take the tactic/maneuver previously determined in the behavior model may be displayed as a green dot. The red line indicates the trajectory of the tracking point, the blue line indicates the trajectory of the actual fighter, and the green dots and the dotted lines connected thereto indicate the relationship between the tracking point and the fighter at each time point. In this way, the processor 110 may generate a tracking point using the steering control model 810 and generate a steering command value for moving along the tracking point.

Referring back to FIG. 8 , the flight control model 820 may determine the movement of the aircraft based on the steering command value. The flight control model 820 may determine the final aircraft movement by performing simulations of the propulsion system, dynamics, flight stability/maneuverability control, and the like. The processor 110 may determine a next state by reflecting the result of the piloting simulation of the flight control model 820 , and the state value for determining the behavior may be updated based on the result of the piloting simulation.

10 shows an overall architecture representing a process by which an apparatus for controlling the steering of an aircraft according to the present disclosure operates.

An apparatus for controlling the operation of an aircraft may obtain input values for action determination from various sensors and process the sensor data to extract feature data that is a action determination state value. In the behavior model 1010, an action tree, fuzzy rules, and machine learning are utilized to determine an action plan based on a manual that reflects what maneuver an aircraft must perform based on feature data to preempt a situation more advantageous than the bandit. In the steering control model 1020, a tracking point indicating the position to which the aircraft will move to perform the maneuvering action determined in the behavior model 1010 is generated, and the stick, throttle, and rudder values, which are steering command values for moving to the tracking point, are derived. can In the flight control model 1030 , the movement of the aircraft may be determined based on the steering command value derived from the steering control model 1020 .

11 illustrates a method of operating a control device according to the present disclosure.

Since each step of the operation method of FIG. 11 may be performed by the control device 100 of FIG. 1 , a description of the content overlapping with that of FIG. 1 will be omitted.

In operation S1110, the control device 100 may acquire at least one of data about the aircraft and data about the environment around the aircraft. In an embodiment, the control device 100 may obtain data about the aircraft (eg, the position of the aircraft, the amount of fuel) and data about the surrounding environment (eg, the position of the bandit) through the sensor. The sensor may include a vision sensor, a RADAR sensor, a GPS/INS sensor, and the like, and the type and quality of data that can be acquired may vary depending on the type of sensor.

In operation S1120, the control device 100 may extract a state value for action determination from the obtained data. In an embodiment, the state value for determining the behavior may be feature data extracted from acquired sensor data, and the feature data includes positional geometry information indicating the relative position between the aircraft and other aircraft, remaining fuel energy, current speed data, etc. may include In an embodiment, the state value for determining the behavior may include at least one of a distance between a plurality of aircraft, a side angle, a nose crossing angle, and an antenna aiming angle.

In operation S1130, the control device 100 may determine the maneuvering behavior of the aircraft from the state value based on the behavior model based on the fuzzy rule and the behavior tree structure. The behavior model can be used to output the optimal maneuvering behavior to favor the battle based on the state value for the behavior judgment obtained earlier. Behavior models can be designed using behavior tree structures and fuzzy rules. The behavior model is based on a behavior tree structure that is modeled based on a manual, and at least one decision process in the behavior model may be implemented according to the fuzzy rule. The action tree means a tree structure designed for action judgment based on the analysis of the manual, and the fuzzy rule expresses the parts vaguely described in the manual (for example, judgment of superiority in the state of the bandit) It is a logical structure that can be used to make decisions under conditions vaguely described in the manual in an action tree structure. An initial model designed using the behavior tree structure and fuzzy rules can be optimized through machine learning, and machine learning can include genetic algorithms as a representative technique of evolutionary computation that mimics the evolution of living things.

In step S1140 , the control device 100 may generate a tracking point corresponding to a maneuvering action. The control device 100 may generate a tracking point for performing the maneuvering behavior determined based on the behavior model above, and may derive a steering command value for controlling the operation of the aircraft according to this.

In operation S1150, the control device 100 may control the operation of the aircraft according to the tracking point. The control device 100 may determine the movement of the aircraft based on the steering command value.

The electronic device or terminal according to the above-described embodiments includes a processor, a memory for storing and executing program data, a permanent storage such as a disk drive, a communication port for communicating with an external device, a touch panel, and a key (key) , a user interface device such as a button, and the like. Methods implemented as software modules or algorithms may be stored on a computer-readable recording medium as computer-readable codes or program instructions executable on the processor. Herein, the computer-readable recording medium includes a magnetic storage medium (eg, read-only memory (ROM), random-access memory (RAM), floppy disk, hard disk, etc.) and an optically readable medium (eg, CD-ROM). ), and DVD (Digital Versatile Disc)). The computer-readable recording medium is distributed among computer systems connected through a network, so that the computer-readable code can be stored and executed in a distributed manner. The medium may be readable by a computer, stored in a memory, and executed on a processor.

This embodiment may be represented by functional block configurations and various processing steps. These functional blocks may be implemented in any number of hardware and/or software configurations that perform specific functions. For example, an embodiment may be an integrated circuit configuration, such as memory, processing, logic, look-up table, etc., capable of executing various functions by the control of one or more microprocessors or other control devices. can be hired Similar to how components may be implemented as software programming or software components, this embodiment includes various algorithms implemented in a combination of data structures, processes, routines or other programming constructs, including C, C++, Java ( Java), assembler, etc. may be implemented in a programming or scripting language. Functional aspects may be implemented in an algorithm running on one or more processors. In addition, the present embodiment may employ the prior art for electronic environment setting, signal processing, and/or data processing. Terms such as “mechanism”, “element”, “means” and “configuration” may be used broadly and are not limited to mechanical and physical configurations. The term may include the meaning of a series of routines of software in association with a processor or the like.

The above-described embodiments are merely examples, and other embodiments may be implemented within the scope of the claims to be described later.

Claims (11)

A method of controlling the operation of an aircraft, comprising:
acquiring at least one of data about the aircraft and data about an environment around the aircraft;
extracting a state value for action determination from the obtained data;
determining the maneuvering behavior of the aircraft from the state value based on a behavior model based on a fuzzy rule and a behavior tree structure;
generating a tracking point corresponding to the maneuvering action; and
and controlling the maneuvering of the aircraft according to the tracking point. According to claim 1,
wherein the behavior model is based on a behavior tree structure modeled based on a manual, and at least one decision process is implemented according to the fuzzy rule in the behavior model. 3. The method of claim 2,
wherein at least one of a member function and a fuzzy variable for the fuzzy rule is optimized through machine learning. 4. The method of claim 3,
wherein the machine learning comprises a genetic algorithm. 4. The method of claim 3,
wherein optimizing the fuzzy variable comprises varying a fuzzy variable affecting the maneuvering behavior. According to claim 1,
Controlling the flight of the aircraft comprises:
generating a steering command value of the aircraft from the tracking point; and
The method of claim 1, comprising the step of performing a piloting simulation of the aircraft based on the steering command value. 7. The method of claim 6,
and the state value is updated based on a result of the piloting simulation. According to claim 1,
The state value includes at least one of a distance between a plurality of aircraft, a side angle, a nose cross angle, and an antenna aiming angle. According to claim 1,
In the determining of the maneuvering behavior of the aircraft, the maneuvering behavior is determined at each preset period, and the period is set differently according to the amount of change in the acquired data. A non-transitory computer-readable storage medium comprising:
A medium configured to store computer readable instructions, comprising:
The computer readable instructions, when executed by a processor, cause the processor to:
acquiring at least one of data about the aircraft and data about the environment around the aircraft;
extracting a state value for action determination from the obtained data;
determining the maneuvering behavior of the aircraft from the state value based on a behavior model based on a fuzzy rule and a behavior tree structure;
generating a tracking point corresponding to the maneuvering action; and
controlling the maneuvering of the aircraft according to the tracking point;
A non-transitory computer-readable storage medium configured to perform a method of controlling the operation of an aircraft. a memory for storing at least one instruction; and
By executing the at least one command, at least one of data about the aircraft and data about the environment around the aircraft is obtained, a state value for action determination is extracted from the obtained data, and a fuzzy rule and an action tree structure are and a processor for determining the maneuvering behavior of the aircraft from the state value based on the behavior model based on the state value, generating a tracking point corresponding to the maneuvering behavior, and controlling the operation of the aircraft according to the tracking point. that is, the aircraft's steering control device.

Images