KR102128551B1 - Apparatus for modeling of reconnaissance behavior and the method thereof - Google Patents

Abstract

A reconnaissance behavior modeling method, according to an embodiment of the present invention, comprises: a first step of generating an environmental agent providing a virtual environment, an unmanned vehicle agent performing a reconnaissance operation in the virtual environment, a manned vehicle agent, and a reconnaissance personnel agent; a second step of allowing the unmanned vehicle agent to select a driving route and to search while moving along the driving route; a third step of allowing the unmanned vehicle agent to select an alighting point on the driving route when an unsearchable area is found while the unmanned vehicle agent is moving, and to transmit the driving route and the alighting point to the manned vehicle agent; a fourth step of allowing the manned vehicle agent to receive the driving route and the alighting point, to move along the driving route, and to make the reconnaissance personnel agent get off at the alighting point; a fifth step of allowing the reconnaissance personnel agent to board the manned vehicle agent after searching the unsearchable area; and a sixth step of storing a simulation result of reconnaissance behaviors according to the second to fifth steps. The second to fifth steps are performed according to a preset rule of conduct.

Description

Apparatus for modeling and method of search behavior {APPARATUS FOR MODELING OF RECONNAISSANCE BEHAVIOR AND THE METHOD THEREOF}

The present invention relates to an agent-based search behavior modeling apparatus and method.

With the recent development of unmanned technology, efforts have been made to increase the efficiency of work by unmanning simple and repeat activities step by step. However, in order for the unmanned system to replace the work (search, reconnaissance, etc.) performed by humans, scientific analysis using simulation techniques is required.

Recently, there is an example of an invention that proposed a search system for missing persons using a cluster drone, but no technique has been devised to analyze their effective operation. In foreign countries, simulation simulation models have been used to conduct search simulations in disaster and terror situations, but due to limitations in simulation capabilities, it is difficult to describe the detailed behavior of search equipment and search personnel.

Specifically, in the case of the prior art, the analysis of the movement/input path for each manned and unmanned search equipment, the range of walking and boarding search, etc. is limited. This is because the prior art is a model based on the Lanchester equation, and when entering a scenario, the analyst must manually input detailed tactical actions. Therefore, there is a limitation in consistently simulating and analyzing detailed search behaviors of the unmarried agent.

In addition, the prior art has limitations in that it is difficult to grasp the reconnaissance area on the route for each agent for search and the movement delay time according to the terrain roughness.

An object of the present invention is to provide a search behavior modeling apparatus and method capable of consistently simulating and analyzing detailed search behaviors of an unmarried agent as being devised to solve the aforementioned problems.

Another object of the present invention is to simulate the search behavior in detail according to the characteristics of the agent and the terrain.

Another object of the present invention is to provide a search behavior modeling apparatus and method capable of establishing an effective search plan using an unattended agent.

The search behavior modeling method according to an embodiment of the present invention for solving the above-described problems generates an environment agent providing a virtual environment, an unmanned vehicle agent performing a search operation in the virtual environment, a manned vehicle agent, and a search personnel agent The first step, the second step of the unmanned vehicle agent selecting a driving route and searching while moving along the driving route, and when the unmanned vehicle agent finds a non-searchable area while moving, get off point on the driving route The third step of selecting, and delivering the driving route and the disembarking point to the manned vehicle agent, after the manned vehicle agent receives the driving route and the disembarking point, after moving along the driving route, the disembarking point Step 4 of unloading the search personnel agent, simulation of the search action according to the fifth step and the second to fifth steps of boarding the manned vehicle agent after the search personnel agent searches for the unsearchable area A sixth step of storing the results is included, and the second to fifth steps are performed according to a predetermined behavior rule.

The preset action rule is characterized in that the action for the search operation is a rule related to at least one action among movement, search, route selection, and drop-off point selection.

In the second to fifth steps, the unmanned vehicle, manned vehicle, and search personnel agent have a predetermined average maneuvering speed, and when moving according to the preset behavior rule, the terrain roughness between the current location and the next location According to the above, it is characterized in that it moves after correcting each preset average starting speed.

In the second and fifth steps, the driverless vehicle and the search personnel agent each have a preset detection range and whether there is a tree topography at the current location when searching at their current location according to the preset behavior rule. According to the above, each of the preset detection ranges is corrected and then searched.

In the second step, the driverless vehicle agent selects a driving route in which a search area is maximized or a search time is minimum according to the preset behavior rule.

The non-discoverable area is an area including a pre-set suspicious area according to the search scenario, and in the third step, the unmanned vehicle agent gets off its current position when it finds the non-discoverable area while moving along the driving route. And a step of adding a drop-off point candidate that moves while being added to the branch candidate and a step of selecting a drop-off point that selects a candidate drop-off point closest to the suspected area from among the drop-off point candidates previously added by the driverless vehicle agent. Should be

The simulation result of the search action is at least among the total search area searched by each of the unmanned vehicle, manned vehicle, and search personnel agent, the total search time spent on each search, information on each movement path, and information related to the environment agent. It is characterized by including one.

The search behavior modeling method further includes a sixth step of analyzing a simulation result of the search behavior according to a preset criterion, wherein the preset criterion includes the search efficiency, the time required for the search, and the unmanned vehicle, the manned vehicle, and the search activity. It characterized in that it comprises at least one of the contribution of each search personnel agent.

A search behavior modeling apparatus according to another embodiment of the present invention simulates a search behavior according to a preset search scenario and provides an environment agent providing the virtual environment, an unmanned vehicle agent performing a search operation in the virtual environment, a manned vehicle agent, and Agent generation unit for generating search personnel agent, unmanned vehicle agent, manned vehicle agent, and control unit for controlling search personnel agent to act according to a predetermined action rule in the virtual environment, information related to the environment agent, and unmanned vehicle agent, manned A simulation unit simulating the search behavior by providing information related to at least one of a vehicle agent and a search personnel agent to an unmanned vehicle agent, a manned vehicle agent, and a search personnel agent, and a search plan storage unit storing simulation results of the search activity Includes.

The driverless vehicle agent selects a driving route, searches while moving along the driving route, and selects a disembarking point on the driving route when it finds an unsearchable area while moving, and uses the driving route and the disembarking point as the simulation unit Forward, the manned vehicle agent receives the driving route and the disembarkation point from the simulation unit, moves along the driving route, and unloads the search agent agent at the disembarking point, and the search agent agent detects the unsearchable area. And after searching, boarding the manned vehicle agent.

The preset action rule is characterized in that the action for the search operation is a rule related to at least one action among movement, search, route selection, and drop-off point selection.

The present invention can consistently simulate and analyze search behaviors using agents that perform detailed search behaviors according to the behavior rules, so that more objective simulation results can be obtained without analyst intervention.

In addition, as the agent performing the search operation according to the behavioral rules of the present invention has a different movement speed and detection range according to the characteristics of the terrain, the accuracy of the simulation result is improved.

In addition, an agent performing a search operation according to the behavioral rules of the present invention selects a disembarkation point for an area where search is not possible, and other agents perform search, and appropriate search activity is provided under a given condition using a mutually cooperated presence agent. Not only can a plan be made, but it can also be used for other purposes, such as planning a disaster/crime response plan.

1 is a diagram for explaining interactions between agents for simulating search behavior according to an embodiment of the present invention.
2 is a diagram showing the configuration of a search behavior modeling apparatus according to an embodiment of the present invention.
3 is a view for specifically describing a process in which an environmental agent is generated.
4 is a flowchart for explaining the operation of the driverless vehicle agent.
5 is a flowchart for explaining the operation of the manned vehicle agent.
6 is a flowchart for explaining the operation of the search personnel agent.
7A and 7B are flowcharts for specifically explaining the route selection of an unmanned vehicle agent according to an action rule.
8 is a flowchart for specifically describing movement of each agent according to an action rule.
9A and 9B are diagrams for specifically explaining the discovery of each agent according to the action rule.
10 is a flowchart for specifically explaining the selection of an unattended vehicle agent's drop-off point according to an action rule.
11 is a view for explaining a simulation result of a search activity.
12 is a flowchart illustrating a search behavior modeling method according to another embodiment of the present invention.

Hereinafter, exemplary embodiments disclosed herein will be described in detail with reference to the accompanying drawings, but the same or similar reference numerals are assigned to the same or similar elements, and overlapping descriptions thereof will be omitted. The suffixes "modules" and "parts" for the components used in the following description are given or mixed only considering the ease of writing the specification, and do not have meanings or roles distinguished from each other in themselves. In addition, in describing the embodiments disclosed in this specification, detailed descriptions of related well-known technologies are omitted when it is determined that they may obscure the gist of the embodiments disclosed herein. In addition, the accompanying drawings are only for easy understanding of the embodiments disclosed in the present specification, and the technical spirit disclosed in the specification is not limited by the accompanying drawings, and all modifications included in the spirit and technical scope of the present invention , It should be understood to include equivalents or substitutes.

The search behavior modeling apparatus and method according to an embodiment of the present invention can consistently simulate and analyze the search behavior, so that more objective simulation results can be obtained without the intervention of an analyst. In addition, the present invention improves the accuracy of the simulation result because the moving speed and the detection range of the agent performing the search operation vary depending on the terrain characteristics.

In addition, the present invention can not only establish an appropriate search activity plan under a given condition by using a mutually exclusive agent, but can also be used for other purposes such as a disaster/crime response plan simulation. The agent refers to an entity performing a search operation in a virtual environment to simulate a search action. It will be described below with reference to the accompanying drawings.

1 is a diagram for explaining interactions between agents for simulating search behavior according to an embodiment of the present invention.

Referring to FIG. 1, in order to simulate a search action according to an embodiment of the present invention, the search action modeling device generates a plurality of agents. The plurality of agents may include an environmental agent (H), an unmanned vehicle agent (U), a manned vehicle agent (G), and a search personnel agent (I).

The environment agent H provides a virtual environment for performing search operations to the remaining agents U, G, and I. The virtual environment is a conversion of the actual map to be suitable for the simulation of the search activity. Details will be described later.

The unmanned vehicle agent U may select a route, move, select a drop-off point, and search according to a predetermined behavior rule for search.

The manned vehicle agent G may move, track unmanned vehicles, stop, and wait according to a predetermined behavior rule for search.

The Search Personnel Agent (I) may travel, route, navigate, board and disembark for search.

In order to simulate the search behavior, the following description will be mainly focused on the content of each agent interacting.

The environment agent H provides a virtual environment to the remaining agents U, G, and I, and the remaining agents U, G, and I performing a search operation act in accordance with a predetermined action rule in the virtual environment while searching. To proceed. As the search operation progresses, the state variable of the environmental agent H is changed, and the state variable is stored separately and included in the search activity simulation result.

The driverless vehicle agent U selects a driving route and searches while moving. The driving route is transmitted to the manned vehicle agent G. If the unmanned vehicle agent U finds an unsearchable area while on the move, the drop-off point is selected and transmitted to the manned vehicle agent G and the search personnel agent I. The unmanned vehicle agent U repeatedly selects a disembarkation point for an area where searching is impossible, and repeats movement, search, and disembarkation point selection until the destination of the driving route is reached.

The manned vehicle agent G moves along the unmanned vehicle agent U using the received driving route. The manned vehicle agent (G) stops so that the search personnel agent (I) can get off when it reaches the previously received drop-off point.

The search personnel agent (I) gets off at the drop-off point and moves to the previously unsearchable area to search. Unlike the vehicle agents (U, G), the search personnel agent (I) can also move to terrain other than roads, so it is also possible to search the area delivered by the unmanned vehicle agent (U) to an unsearchable area. The search personnel agent (I) returns to the manned vehicle agent (G) stationary after searching all the above unsearchable areas and boards.

The manned vehicle agent G moves along the driving route to the next drop-off point when the search personnel agent I has all boarded.

Each agent (U, G, I) state variable that is changed as each agent (U, G, I) acts for a search operation according to a predetermined action rule is stored separately and included in a simulation result of the search action.

The unmanned vehicle agent U may be searched while traveling along the driving route selected by the driver, and the unsearchable area may be searched for by the unmanned vehicle agent G disembarking the search personnel agent I at the drop-off point.

As described above, the present invention can simulate a search activity using agents (U, G, I) that cooperate with each other in a virtual environment, and thus, an appropriate search activity plan can be established, and other disaster and crime response plans can be established. It can be used for purposes.

The above is based on the operation of each agent (U, G, I) interacting with the simulation of the search activity. Hereinafter, a search behavior modeling apparatus corresponding to a physical configuration for simulating a search behavior will be described.

2 is a diagram showing the configuration of a search behavior modeling apparatus according to an embodiment of the present invention.

The search behavior modeling apparatus 100 may simulate a search behavior according to a preset search scenario in a virtual environment.

The preset search scenario may include information on a search target area, information on a start and end point of a search operation, and information on a suspect area as information for creating a virtual environment. The suspicious area is an area included in the search target area and means a virtual area that must be searched when performing a search operation because a threat is expected to exist.

The preset search scenario may further include information regarding a search target. An example of the search target is a maximum search area or a minimum search time. However, it is not limited thereto, and the search target may be specifically set by adding various conditions to establish an appropriate search plan.

Referring to FIG. 2, the search behavior modeling apparatus 100 according to an embodiment of the present invention includes an agent generation unit 110, a control unit 120, a mock unit 130, a search method storage unit 140, and a search method The analysis unit 150 may be included.

The agent generation unit 110 generates a presence or absence agent (H, U, G, I) to simulate the search operation. Specifically, the agent generating unit 110 generates the environment agent H, the unmanned vehicle agent U, the manned vehicle agent G, and the search personnel agent I according to the search scenario.

The agent generator 110 creates an environment agent H by modeling the terrain to be searched according to the search scenario. The generated environment agent H provides the virtual environment by interworking with the remaining agents U, G, and I.

According to an embodiment, the agent generator 110 may divide the entire area of the terrain to be searched into a grid of 100x100 by applying a grid-based map design technique for the modeling. The grid map corresponding to the virtual environment uses a patch (1x1 size grid) unit and follows the p[i,j] notation. Each patch has initial values for terrain type (road, tree, suspect area), latitude and longitude coordinates, altitude, search area, and patch size, as shown in Table 1 below. The Patch Size variable, which represents the patch size, uses the large-scale distance calculation method to indicate the distance on the actual terrain.

variable Initial value Terrain type
(Terrain type) - road, forest, ground, ROI (road, tree, land, suspicious area) Geographical Location
(Geographical location information) - [Longitude, latitude, altitude] (longitude, latitude, altitude) ReconTargetState
(Search or not) - True / False (True/False) Patch size
(Patch size) PS Great-Circle Distance

In Table 1, the terrain type has values corresponding to one of roads, trees, land, and suspect areas. Geographic location information includes information about longitude, latitude and altitude. Each patch has a value of True/False depending on whether it is searched or not. In the case of the patch size, the unit is PS and has a value according to the calculation of the large winding distance.

The large distance is the shortest distance between two points on the earth, and the distance between the two points and the two points measured on the circumference (the large circle) that occurs when the plane passing through the center of the earth meets the surface of the earth, is calculated as in Equation 1 below.

,

과

,

는 임의의 패치 A와 B의 위도 및 경도를 라디안으로 변환한 값을 나타내며, 산출된

에 평균 지구 반경

을 곱하여 최종 거리를 계산한다. 이를 기반으로 패치크기는 격자지도에서 상하, 좌우, 대각선 방향의 패치간 거리의 평균값으로 설정한다. In Equation 1 above

,

and

,

Denotes the latitude and longitude of arbitrary patches A and B converted to radians, and is calculated.

Average Earth Radius

Multiply by to calculate the final distance. Based on this, the patch size is set as the average value of the distance between the patches in the vertical, horizontal, and diagonal directions on the grid map.

The virtual environment that the environment agent H provides to the agents U, G, and I performing a search operation is a variable value according to the position of each agent U, G, I in the grid map. Specifically, the environment agent H may transmit at least one information on the terrain type, the geographic location information, the search area, and the patch size according to the location of each agent U, G, and I.

The environmental agent H has a state variable of a grid map on which a search operation is performed. State variables according to an embodiment are shown in Table 2 below.

Terrain type Variable type Variable value ALL ReconState (Search or not) True / False VisitState (visit or not) True / False VisitAgent U, G, I VisitSequence U0, U1, G0, G1 ,I0, I1... Road dismountHere (Where to get on and off) True / False ROI ROI_ID (suspect zone identifier) (0, 1, 2...) ClearState (Completion of search for suspected areas) True / False

The environmental agent H records the status of the search operations of the agents U, G, and I using the above state variable. Specifically, the environmental agent (H) is interworked with each agent (U, G, I) based on the location of the agent (U) in the grid map to search for (eg True), whether to visit (eg True), visited agents (for example, U), visit order by agent (for example, U2), check-in and drop-off points (for example, True), and whether to search for suspected areas (for example, False) are continuously updated.

The state variable recorded by the environmental agent H is stored in the search room storage unit 140, which will be described later.

The agent generation unit 110 models an object performing a search action according to a predetermined behavior rule as a presence/absence agent and generates it. Specifically, the agent generator 110 may generate at least one unmanned vehicle agent U, a manned vehicle agent G, and a search personnel agent I according to the search scenario. According to the search scenario, each agent (U, G, I) may be generated in plural. Hereinafter, for convenience, each agent (U, G, I) will be described on the assumption that one is generated.

According to an embodiment, the agent generator 110 may model the presence or absence agent using the initial values shown in Table 3 below.

variable Initial value real Virtual environment (lattice map) Initial position
(Initial location)

Geodetic p[i,j] Average maneuvering speed
(Average speed)

70km/h 70km/h

60km/h 60km/h

4.5km/h 4.5km/h Detection distance
(

)

3 km (

) patches

1.5 km (

) patches

1.5 km (

) patches Visible range
(Field Of View) FOV 360 degrees 360 degrees

는 각 에이전트의 초기 위치에 관한 정보로서 실 지형 환경에서의 경위도 좌표와 가장 가까운 속성을 갖는 격자 지도상 좌표 p[i,j]이다.평균 기동 속력(Average speed)은 수색 작업을 고려하여 가정한 속력이다. 구체적으로 무인 차량, 유인 차량 에이전트(U, G)는 도로 위에서 이동 시의 속력이고, 수색 인원 에이전트(I)는 평지에서 보행 시의 속력이다. 수색 인원 에이전트(I)가 유인 차량 에이전트(G)에 탑승하여 이동하는 경우에는 유인 차량 에이전트(G)의 평균 기동 속력을 따른다.In Table 3 above

Is information about the initial location of each agent, and is the coordinate p[i,j] on the grid map with properties closest to the latitude and longitude coordinates in the real-world environment. The average speed is assumed by considering the search operation. Speed. Specifically, unmanned vehicles, manned vehicle agents U and G are speeds when moving on the road, and search personnel agents I are speeds when walking on flat land. When the search personnel agent I moves on board the manned vehicle agent G, the average maneuvering speed of the manned vehicle agent G is followed.

)도 그에 따라 상이하다. 구체적으로 최대 탐지 거리(

)의 단위는 각 에이전트(U, G, I)의 위치에서 직선상 탐지 거리가 확보된 패치의 개수이다. 가시 범위(FOV)는 모든 에이전트가 동일하게 360도이다. 일 실시예에 따라 별도의 관측 센서를 탑재하지 않은 차량 에이전트(G)의 경우 탑승하고 있는 수색 인원 에이전트(I)의 최대 탐지 거리(

)를 적용할 수 있다.It is assumed that the search is performed through the observation sensors mounted on the vehicle agents U and G and the naked eye of the search agent I. Therefore, the maximum detection distance (

) Is different accordingly. Specifically, the maximum detection distance (

The unit of) is the number of patches in which a linear detection distance is secured at the location of each agent (U, G, I). The visible range (FOV) is 360 degrees for all agents equally. In the case of the vehicle agent G without a separate observation sensor according to an embodiment, the maximum detection distance of the search personnel agent I on board (

) Can be applied.

Each agent (U, G, I) created has its own state variable. Table 4 shows the state variables of the agents U, G, and I according to an embodiment.

agent Variable type Variable value type ALL Current speed
(current speed)

-km/h Current location
(current location)

p[i,j] Current travel time
(current moving time)

real time Total search time
(total recon time)

real time Detection range
(detection range)

set of patches Next location
(next location)

p[i,j] Middle destination
(temp destination)

p[i,j] Final destination
(final destination)

p[i,j] Search Personnel (I) Get off
(dismountState) - dismount/ boarding Total boarding time
(total boarding time)

real time Manned Vehicle (G) Total stop time
(total stop time)

real time

Each agent (U, G, I) continuously updates its status variable while performing a search operation. In the case of the search personnel agent (I), it is possible to board or get off the manned vehicle agent (G). It has status variables related to whether to get off the train and the total boarding time. Specifically, the total boarding time increases only when the search personnel agent (I) is boarding, and this time is used to calculate the total recon time of the search personnel agent (I). do.

In the case of the manned vehicle agent G, the search personnel agent I should be disembarked at the drop-off point and stopped until the search personnel agent I finished searching and boarding. Therefore, the total stop time increases from the time when the manned vehicle agent G stops at the drop-off point to departs toward the next drop-off point.

The status variables of each agent U, G, and I are stored in the search method storage unit 140, which will be described later, after the row for the search operation ends.

As described above, the agent generation unit 110 models and generates unattended agents (H, G, U, and I) having detailed state variables in connection with the search activity, and each agent (U, which performs the search operation) Search behavior can be simulated in detail according to the characteristics of G and I).

Agents (U, G, and I) for search may perform at least one of movement, search, route selection, and drop-off point selection, which are actions for the search operation, in the virtual environment according to a predetermined action rule.

The preset behavior rules are the basis for the behavior of each agent (U, G, I) and are a set of rules for simulating the behavior and state changes according to the environmental conditions recognized by each agent (U, G, I). . The preset action rule may be a rule related to at least one action among movement, search, route selection, and drop-off point selection, which are actions for the search operation.

The controller 120 may control the unmanned vehicle agent U, the manned vehicle agent G, and the search personnel agent I to act according to a predetermined behavior rule in the virtual environment. The controller 120 may control movement, navigation, route selection, and drop-off point selection, which are actions of each agent U, G, and I.

In relation to the movement for the search activity, the control unit 120 is configured to control the status of each agent U, G, or I when each agent U, G or I moves from the current location to the target location. Each average maneuvering speed is controlled to move to the corrected current speed considering the terrain roughness.

Specifically, the controller 120 calculates the slope of the terrain and corrects the average starting speed of each agent (U, G, I) by using the steepness according to the slope, and adjusts the corrected speed as the current speed. Agent (U, G, I). This is to simulate that the higher the slope, the slower the speed.

) 내에 수목 지형이 존재하는 경우 상기 탐지 범위(

)가 감쇠되도록 보정하여 해당 에이전트(U, G, I)에 전달한다. 그 후 에이전트(U, G, I)가 해당 영역에 대해 탐색을 수행하고 감쇠된 탐지 범위(

)에 포함된 영역(패치)만이 탐색된다(수색 여부(ReconState)가 True로 변경). 이는 수목 지형에서는 평지와 달리 가시거리가 줄어드는 것을 모의하기 위함이다.The control unit 120 detects the detection range from the current location of the agents U, G, and I in relation to the search for the search activity (

If there is a tree topography within the above detection range (

) Is corrected to be attenuated and transmitted to the corresponding agents (U, G, I). The agent (U, G, I) then performs a search on the area and the attenuated detection range (

), only the area (patch) included is searched (ReconState is changed to True). This is to simulate that the visibility distance is reduced, unlike flat land, in the tree topography.

The control unit 120 relates to a route selection for a search activity, and the unmanned vehicle agent G selects a driving route in which the search area is maximized or the total search time is minimum from the start point to the end point according to the search scenario. Control to select. This is to simulate a search action according to the search target of the search scenario.

The control unit 120 controls to select a point closest to a non-searchable area from among candidates for alighting points added by an unmanned vehicle agent U in relation to selection of alighting points for a search action. The non-searchable area is an area including a suspected area according to the search scenario and means an area not searched despite the search of the unmanned vehicle agent U.

When the unmanned vehicle agent U finds the unsearchable area while moving along the driving route selected by the unmanned vehicle agent U, the controller 120 controls to add the current location to the candidate point of disembarkation, and the unmanned vehicle agent U If it is outside the preset radius from the non-searchable area, it is controlled to select a candidate drop-off point closest to the non-searchable area as at least one of the at least one drop-off point candidates already added.

According to an embodiment, the control unit 120 may be implemented as a memory device including commands related to the action rules of the agents U, G, and I.

The controller 120 is configured to simulate autonomously determining and searching according to the environment recognized by each agent U, G, and I in the virtual environment, and the role of the controller 120 is through the agent generator 110 Can be replaced.

Specifically, according to an embodiment, the agent generator 110 may generate an agent (U, G, I) including a command related to the action rule of each agent (U, G, I). In this case, the search behavior modeling apparatus 100 does not include the control unit 120, and each agent U, G, and I performs an action for a search operation solely according to its own action rules. Each agent (U, G, I) can perform calibration of movement speed, attenuation of detection range, route selection and drop-off point selection alone.

As described above, the agents (U, G, and I) that perform a search operation according to a predetermined action rule have different movement speeds and detection ranges according to the characteristics of the terrain, and thus the accuracy of the simulation results is improved.

In addition, according to the present invention, each agent (U, G, I) conducts a search operation in cooperation with each other by selecting a route considering a search area or a total search time, and selecting a disembarkation point for a point where search is impossible. Not only can you make a plan, it can also be used for other purposes, such as planning a disaster/crime response plan.

The mock unit 130 simulates the search behavior of the agents U, G, and I in a virtual environment.

Specifically, the simulation unit 130 transmits related information so that the environment agent H, the unmanned vehicle agent U, the manned vehicle agent G, and the search personnel agent I can interact with each other. To this end, the simulation unit 130 provides at least one of the state variables of each agent H, U, G, and I to the remaining agents H, U, G, and I, respectively.

As an example, the mock unit 130 receives the unmanned vehicle agent U when the driving route or the getting-off point is received and provides it to the environment agent H and the manned vehicle agent G. In this specification, that each agent (U, G, I) is described as recognizing the virtual environment provided by the environment agent H is the same as meaning that it obtains the state variable of the environment agent H delivered by the mock unit 130. Do.

The simulation unit 130 generates road network information including a plurality of intersection points existing between a start point and an end point according to the search scenario in the virtual environment grid map, and controls the controller 120 and the driverless vehicle agent ( U).

The control unit 120 transmits information related to a search area or a search time between each point from the start point to the end point using the road network information to the unmanned vehicle agent U, and the unmanned vehicle agent U is The point at which the search area becomes the maximum or the search time becomes the minimum is selected as the next position. More details will be described later.

The search room storage unit 140 stores the simulation result of the search action.

Specifically, the search method storage unit 140 receives and stores the state variables of all the agents (H, U, G, I) whenever the action for the search operation of each agent (U, G, I) ends. . Therefore, the simulation result of the search activity includes information of the grid map and the state variable of the environment agent H, and at least one of the state variables of the unmanned vehicle agent U, manned vehicle agent G, and search personnel agent I The state variable of can be included.

According to an embodiment, the simulation result of the search activity includes information related to the environmental agent H and at least one of the total search area, total search time, and movement path information of each agent U, G, and I. It may include.

According to an embodiment, the search room storage unit 140 may include a display unit. The search method storage unit 140 displays a grid map provided by the environmental agent H on the display unit, and is associated with a search activity such as the location of each agent U, G, I, movement path, etc. on the grid map. Information can be displayed in real time.

The search plan analysis unit 150 analyzes the simulation results of the stored search behavior according to a preset criterion. The preset criteria may include at least one of search efficiency, search time, and contribution of each agent (U, G, I) to search activity.

The search plan analysis unit 150 calculates search efficiency as shown in Equation 2 below using information about the search target area included in the search scenario and the total search area.

(Area of Operation)는 상기 수색 대상 면적에 속하는 패치들의 집합이다. In Equation 2 above

(Area of Operation) is a set of patches belonging to the search target area.

(Area of Recon)은 상기 수색 대상면적에서 탐색이 완료된 면적에 속한 패치들의 집합으로서, 각 에이전트(U, G, I)의 총 탐색 면적을 모두 합한 것이다.

(Area of Recon) is a set of patches belonging to the area where the search is completed in the search target area, and is the sum of the total search areas of each agent (U, G, I).

,

,

및

)을 이용하여 상기 탐색 효율을 계산한다.The search plan analysis unit 150 is a set of patches for forest and suspected areas in each set (

,

,

And

) To calculate the search efficiency.

는 수목 지형 탐색에 대한 가중치이고

는 의심 구역 탐색에 대한 가중치이다. 의심 구역의 탐색은 수목 지형의 탐색보다 중요도가 높으므로

는

보다 큰 값을 가질 수 있다. 구체적으로 예를 들면,

는 0.4이고

는 0.6일 수 있다. 다만 이에 한정되는 것은 아니며 상기 각 가중치(

,

)는 상기 수색 시나리오에 따른 수색 대상 면적 및 수색 목표를 고려하여 상기 수색 행위의 모의 결과를 분석하기에 적절한 값으로 각각 설정될 수 있다.

Is the weight for exploring tree topography

Is the weight for searching the suspected area. The search for suspected areas is more important than the search for tree topography,

The

It can have a larger value. Specifically, for example,

Is 0.4

May be 0.6. However, the present invention is not limited thereto, and the respective weights (

,

) May be respectively set to a value suitable for analyzing the simulation result of the search action in consideration of the search target area and the search target according to the search scenario.

The search time required refers to the time from the time when the agents U, G, and I generated according to the search scenario start the search operation to the time when the search operations of all the agents U, G, and I are finished.

)을 더하여 상기 탐색 소요시간을 계산한다.Search plan analysis unit 150 is the total search time of the unmanned vehicle agent (U) and search personnel agent (I) (

) To calculate the search time.

The contribution ratio (CR) refers to the degree to which the unmanned vehicle agent U and the search personnel agent I each contributed to the search activity, and the search plan analysis unit 150 compares each agent (U, with respect to the search target area). Based on the total search area searched by I), it is calculated as in Equation 3 below.

는 상기 수색 대상 면적 중에서 에이전트(U 또는 I)가 탐색한 수목 지형 면적에 속하는 패치들의 집합이고

는 의심 구역의 면적에 속하는 패치들의 집합이다.In Equation 3, E means an agent and is an unmanned vehicle agent U or a search personnel agent I.

Is a set of patches belonging to the tree terrain area that the agent (U or I) has searched among the search target areas.

Is a set of patches belonging to the area of the suspect area.

As described above, the search plan analysis unit 150 analyzes a simulation result of a search activity according to a preset criterion, and can compare and analyze a plurality of simulation results to establish a more appropriate search plan.

Hereinafter, creation of environment agents H, U, G, and I and operations of each agent U, G, I in a virtual environment will be described in more detail with reference to the drawings.

3 is a view for specifically describing a process in which an environmental agent is generated.

The agent generation unit 110 generates an environment agent H by modeling a map of a search target, and a virtual environment (lattice map) provided by the environment agent H includes forest, road, and suspect areas. (Threat) and geodetic information such as latitude and longitude (Geodetic & Altitude etc) are included. 3(a) shows a grid map provided by the generated environmental agent H.

According to one embodiment, the agent generating unit 110 may use the digital elevation data (DTED) and the terrain feature data (FDB) provided by the National Intelligence Service as the terrain data of the search target for generating the environmental agent (H). have. 3(b) shows a process in which the agent generator 110 preprocesses a series of data for grid-type map representation.

The agent generator 110 converts the format of the collected terrain data into a raster format and sets the resolution to 100x100 (1 to 3). When the converted input data is output in xyz format using the Geospatial Data Analysis Library (GDAL) library, the longitude, latitude, and altitude values of each cell are extracted (4). The xyz data output from the FDB data is interpreted as the existence of the terrain if it has a value greater than 0 with reference to the z value.

4 is a flowchart for explaining the operation of the driverless vehicle agent.

4, the driverless vehicle agent U updates its state variable (S41). The state variable of the driverless vehicle agent U to be updated is also transmitted to the search room storage unit 140. The driverless vehicle agent U recognizes the environment at its current location in the virtual environment provided by the environment agent H (S42), and determines the next target location (S43). At this time, the target location (next location) is determined according to a predetermined action rule.

According to an embodiment, the unmanned vehicle agent U may determine the target location using the information transmitted from the controller 120. According to another embodiment, the driverless vehicle agent U may directly determine the next location.

7 will be described in more detail with reference to route selection.

The driverless vehicle agent U determines the speed to the target location.

Specifically, the controller 120 calculates the inclination between the current location of the unmanned vehicle agent U and the target location and uses the roughness accordingly to average the moving speed of the unmanned vehicle agent U Correct (Average speed). According to another embodiment, the driverless vehicle agent U may directly determine the speed up to the target location according to a preset behavior rule.

)를 감쇠시킨다.The unmanned vehicle agent U moves to the target location at the determined speed, and then recognizes the environment at the current location to determine the detection range (S45). Specifically, the control unit 120 detects the range of the unmanned vehicle agent U according to the presence or absence of tree topography (

) Attenuates.

)를 직접 결정할 수 있다.According to an embodiment, the unmanned vehicle agent U may detect its detection range according to a preset behavior rule (

) Directly.

)에 따라 탐색을 한다. 탐지 범위(

)에 포함된 격자 지도상의 패치의 수색 여부(ReconState)의 값은 true로 변경된다.The driverless vehicle agent (U) has a determined detection range (

). Detection range (

The value of whether the patch on the grid map included in) is searched (ReconState) is changed to true.

) 내에 의심 구역이 포함되는 것을 파악하였으나, 탐지 범위(

)가 감쇠됨에 따라 탐색이 완료되지 않은 경우 현재 위치(current location)를 하차 지점으로 선정할 수 있다.The driverless vehicle agent U determines whether to select a current location as a drop-off point after completing the search (S46). Specifically, the unmanned vehicle agent (U) has a detection distance (

), but the scope of detection (

If the search is not completed due to attenuation), the current location can be selected as a drop-off point.

)를 벗어났을 때 기 추가된 하차 지점 후보지 중에서 상기 해당 의심 구역과 가장 인접한 후보지를 하차 지점으로 선정할 수 있다.According to an embodiment, the unmanned vehicle agent U does not immediately select a disembarkation point, but includes the current location in the disembarkation point candidate and the controller 120 determines that the suspected area is the unmanned vehicle agent U. Detection distance (

), the candidate point closest to the relevant suspect area can be selected as the drop-off point among the candidate points of drop-off point already added.

According to an embodiment, the unmanned vehicle agent U may select the disembarkation point from the pre-added disembarkation point candidates according to a preset behavior rule.

The driverless vehicle agent U decides whether to select a drop-off point and recognizes the virtual environment again to continue moving. The unmanned vehicle agent U repeats S41 to S46 until it reaches the end point according to the search scenario.

5 is a flowchart for explaining the operation of the manned vehicle agent. Details overlapping with the operation of the unmanned vehicle agent U are omitted.

Referring to FIG. 5, the manned vehicle agent G updates its state variable (S51 ). The status variable of the updated manned vehicle agent G is also transmitted to the search room storage 140.

The manned vehicle agent G recognizes the environment at its current location in the virtual environment provided by the environment agent H and delivers the driving path of the unmanned vehicle agent U through the mock unit 130. Receive (S52).

The manned vehicle agent G determines a target point for tracking the unmanned vehicle agent U along the received travel route (S53), determines the speed to the target point (S54), and then determines the target according to the determined speed Move to the point.

)를 결정하고(S55), 결정된 탐지 범위(

)에 따라 탐색을 한다. The manned vehicle agent G recognizes the environment at the current location after moving to the target point and detects the detection range (

) (S55), and the determined detection range (

).

Since the main search for the search target area is performed by the unmanned vehicle agent U and the search personnel agent I, the manned vehicle agent G according to an embodiment performs the operation of the next step without performing step S55. Can.

After completing the search, the manned vehicle agent G recognizes the environment at the current location and determines whether the current location corresponds to the drop-off point previously received from the unmanned vehicle agent U. It is judged (S56).

When the current location corresponds to the disembarkation point, the manned vehicle agent G stops for the disembarkation of the search personnel agent I on board and then the disembarked search personnel agent I searches. Wait until you are finished and board again.

When the current location is not the drop-off point or the search personnel agent I has finished boarding, the manned vehicle agent G recognizes the virtual environment to continue moving.

The manned vehicle agent G repeats the above steps S51 to S56 until the unmanned vehicle agent U reaches the last pick-up point.

6 is a flowchart for explaining the operation of the search personnel agent. Details overlapping with unmanned vehicles and manned vehicle agents U and G are omitted.

Referring to FIG. 6, when the manned vehicle agent G stops at the drop-off point, the search personnel agent I gets off at the manned vehicle agent G to update its state variable (S61 ). The status variable of the updated search agent agent I is also transmitted to the search room storage 140.

The search personnel agent I recognizes the environment at its current location in the virtual environment provided by the environment agent H (S62), and searches for the location of the suspect area (threatening terrain) adjacent to it. (S63).

Search personnel agent (I) determines the target location (next location) in order to move to the suspected area (threatening terrain) found, and determines the speed to the target location (next location) according to a predetermined action rule (S65).

)를 결정하고(S66), 결정된 탐지 범위(

)에 따라 탐색을 한다.The search personnel agent (I) recognizes the environment at the current location after moving to the target location, and detects its detection range according to the preset behavior rules (

) (S66), and the determined detection range (

).

The search personnel agent (I) completes the search and recognizes the environment to determine whether all of the search for the suspected area (threatening terrain) has been completed (S67). When all of the search for the suspected area (threatening terrain) is completed, the search personnel agent I moves to the waiting manned vehicle agent G and boards.

When the search of the suspected area (threatening terrain) is not completed, the search personnel agent I repeats S64 to S67. Even at this time, the search agent agent (I) updates his state variable every time.

The search personnel agent I repeats S61 to S67 each time the manned vehicle agent G stops at the drop-off point.

7A and 7B are flowcharts for specifically explaining the route selection of an unmanned vehicle agent according to an action rule.

The driverless vehicle agent U selects a driving route in consideration of the search target according to the search scenario. The road network information generated by the simulation unit 130 is used so that the unmanned vehicle agent U can receive assistance in planning the route to the end point. The road network information is an abstract representation of the main maneuvering path of the unmanned vehicle agent U, and includes information about a plurality of intersection points existing between the start point and the end point according to the search scenario in a grid map that is a virtual environment. Doing.

7A shows an example of the path network.

)인 노드 ①과 종료 위치(

)인 노드 ⑨를 제외한 노드들은 작전지역 내 주요 교차로를 나타내고, 모의부(130)는 각 교차로 사이에 이동 가능한 경로가 존재할 경우 링크(Link)를 생성한다. 이때 상기 링크는 임무계획 수립 시, 우회를 포함한 경로 선정 가능성을 고려하여 양방향으로 생성된다. Referring to Figure 7a, the starting position (

) In node ① and end position (

Nodes except node ⑨,) represent the main intersection in the operation area, and the mock unit 130 creates a link when there is a movable path between each intersection. At this time, the link is created in both directions in consideration of the possibility of route selection including bypass when establishing a mission plan.

The unmanned vehicle agent U plans a route using a greedy algorithm-based global route planning method based on the road network information received from the mock unit 130. Specifically, the unmanned vehicle agent U determines the next moving direction by referring to the cost w of the connected link whenever it reaches the intersection. Links have two cost attributes depending on search area and travel time. The search area cost represents the searchable area when moving to the corresponding link, and the search time cost represents the time required to move to the node opposite to the selected link.

7B is a flowchart for specifically explaining a process of selecting a route by an unmanned vehicle agent U using the path network information.

는 무인 차량 에이전트(U)를 의미한다. Referring to Figure 7b,

Means unmanned vehicle agent U.

)에서 자신과 가장 인접한 교차로 노드(intersection node)를 중간 목적지(

)로 업데이트 한다.In S700, the driverless vehicle agent U is the starting position (

), the intersection node closest to itself is the intermediate destination (

).

)이 중간 목적지(

)를 향하도록 업데이트 한다.In S701, the driverless vehicle agent U is moving direction (

) Is an intermediate destination (

).

=180)내에 포함되는 패치를 인식한다. 구체적으로 예를 들면, 현재 위치에서 무인 차량 에이전트(U)가 북쪽 방향을 바라보고 있는 경우 상기 반원에 포함되어 인식되는 패치는 하기 표 5와 같다.In S702, the driverless vehicle agent U has a radius of 1.5 from the current position (that is, 1.5 patch intervals) and a semicircle facing the direction of movement (

=180). Specifically, for example, when the driverless vehicle agent U in the current position is looking toward the north, patches included in the semicircle and recognized are shown in Table 5 below.

Patch 2
(include) Patch 1
(include) Patch 2
(include) Patch 1
(include) Current location
(U) Patch 1
(include) Patch 2 Patch 1 Patch 2

A total of 5 patches are recognized in Table 5 above. Three patches among the patch 1 located on the top, bottom, left, and right of the unmanned vehicle agent U within the range of one patch interval from the current position are recognized, and the patch located in the diagonal direction of the unmanned vehicle agent U within the patch gap 1.5 range Two of the two are recognized. In S703, the driverless vehicle agent U checks whether there is a patch of a road terrain that has not been searched among the recognized patches.

)가 존재하지 않는 것으로 판단한 경우, 상기 반원의 중심각(

)이 360도 보다 작을 때 S705에서 무인 차량 에이전트(U)는 10도만큼 반원의 중심각(

)을 넓힌 뒤(S706) 도로 지형을 탐색한다(S701-S703). 상기 반원의 중심각(

)이 360도인 경우 현재 위치에서 인접하는 8 방향의 패치가 모두 도로 지형이 아닌 것이므로 무인 차량 에이전트(U)는 이전 위치(

)로 기 설정된 행동 규칙에 따라 이동한다(S707). 이후 해당 위치(

)에서 도로 지형을 탐색한다(S701-S703).The patch of the road terrain among the patches recognized by the driverless vehicle agent U in S704 (

), the center angle of the semicircle (

) Is less than 360 degrees, the unmanned vehicle agent (U) in S705 has a center angle of a semicircle by 10 degrees (

) (S706) and then explore the road topography (S701-S703). Center angle of the semicircle (

) Is 360 degrees, the unattended vehicle agent (U) moves to the previous location (

) Is moved according to a predetermined action rule (S707). After that location (

) To search for the road topography (S701-S703).

)가 존재하는 것으로 판단한 경우, 무인 차량 에이전트(U)는 도로(road) 지형의 패치(

)의 수를 체크한다(S715).The patch of the road terrain among the patches recognized by the driverless vehicle agent U in S704 (

If it is determined that exists, the driverless vehicle agent U is a patch of the road terrain (

) Is checked (S715).

)만이 존재하는 경우 무인 차량 에이전트(U)는 다음 위치(next location,

)를 해당 도로 지형 패치(

)로 업데이트 한다. S728에서 무인 차량 에이전트(U)는 현재 위치(current location,

)에서 상기 다음 위치(next location,

)로 기 설정된 행동 규칙에 따라 이동한다.One road terrain patch in S716(

If only) is present, the driverless vehicle agent U is next location,

) The corresponding road terrain patch (

). In S728, the driverless vehicle agent U is the current location,

) At the next location (next location,

).

)가 존재하는 것으로 판단한 경우, 중간 목적지(

)에 가장 인접한 도로 지형 패치(

)를 선택하여(S717) 해당 도로 지형 패치(

)를 다음 위치인

으로 업데이트 한다(S718). 무인 차량 에이전트(U)는 업데이트된

을 기준으로 S728 단계를 수행한다.In S716, the unmanned vehicle agent (U) patches one or more road terrains (

), the intermediate destination (

) The terrain patch that is closest to the road (

) (S717) to patch the corresponding road terrain (

) Next position

Update to (S718). Unmanned Vehicle Agent (U) has been updated

Based on the step S728 is performed.

이 종료 위치(

)인지 여부를 판단하고, S730에서 종료 위치(

)인 경우 무인 차량 에이전트(U)는 수색 작업을 종료한다. S730에서 종료 위치(

)가 아닌 것으로 판단한 경우 무인 차량 에이전트(U)는 S731에서

이 교차로(intersection)에 해당하는 지 체크한다.In S729, the driverless vehicle agent (U) is

This end position (

), and the end position (S730)

), the driverless vehicle agent U ends the search operation. End position at S730(

), the driverless vehicle agent (U) in S731

Check if it corresponds to this intersection.

이 교차로(intersection)에 해당하는 것으로 판단한 경우 무인 차량 에이전트(U)는 주행 경로를 결정하기 위해

와 연결되는 교차로(intersection) 중에서 각각의 링크의 비용 속성과 수색 목표에 해당하는 경로 선정 기준(Planning criteria)을 고려하여 탐색 면적이 최대가 되거나 탐색 시간이 최소가 되는 교차로(intersection)를 선택하여 중간 목적지(

)로 업데이트 한다(S733, S734). 이후 무인 차량 에이전트(U)는 종료 지점(

)에 도착할 때까지 S701부터 다시 반복하여 각 단계를 수행한다.From S732

If it is determined that this is an intersection, the driverless vehicle agent U may determine the driving route.

Among the intersections that are connected to the middle, select the intersection that maximizes the search area or minimizes the search time by considering the cost attribute of each link and the planning criteria corresponding to the search target. destination(

) (S733, S734). After that, the driverless vehicle agent U will end

Repeat each step from S701 until it reaches ).

)에 해당하는 상기 교차로(intersection)의 정보를 무인 차량 에이전트(U)에 전달함으로써 중간 목적지(

)가 업데이트 될 수 있다.According to an embodiment, the control unit 120 may provide an intermediate destination (

) By passing the information of the intersection corresponding to the unmanned vehicle agent (U) to the intermediate destination (

) Can be updated.

이 교차로(intersection)가 아닌 것으로 판단한 경우 종료 지점(

)에 도착할 때까지 S701부터 각 단계를 수행한다.In S732, the driverless vehicle agent (U)

If it is determined that this is not an intersection, the end point (

Each step is performed from S701 until arrival at.

8 is a flowchart for specifically describing movement of each agent according to an action rule.

)와 다음 위치(

)간의 거리 D를 계산한다.8, in S800, each agent E (U, G, I) is the current location (

) And next position (

Calculate the distance D between ).

)와 다음 위치(

)간의 경사도(slope)를 계산한다. 상기 경사도(slope)는 하기 수학식 4를 이용하여 계산된다.In S801-S802, each agent E (U, G, I) determines the roughness level based on forest density and slope. Each agent E (U, G, I) has its current location (

) And next position (

Calculate the slope between ). The slope is calculated using Equation 4 below.

)와 다음 위치(

)간의 표고차를 나타낸다. 각 에이전트 E(U, G, I)는 모의부(130)를 통해 전달 받는 격자 지도의 상태 변수로부터 상기 표고차 Ht를 구할 수 있다.In Equation 4, Ht is the current position (

) And next position (

). Each agent E (U, G, I) may obtain the elevation difference Ht from the state variable of the grid map received through the simulation unit 130.

)와 다음 위치(

)간 수목 지형의 밀집 정도를 나타내는 값이다. 상기 수색 시나리오에 따라 임의의 수목 밀도가 설정되는 것으로 가정한다. 상기 수목 밀도와 상기 경사도에 따른 험준도의 일 예를 하기 표 6에 정리하였다.For tree density above, current location (

) And next position (

) This is a value indicating the degree of denseness of the tree topography. It is assumed that an arbitrary tree density is set according to the search scenario. Table 6 shows an example of the steepness according to the tree density and the slope.

slope
Tree density 15% or less 16% or more and 45% or less 46% or more 30% or less ground forest Ⅰ forest Ⅱ 31% or more and 60% or less forest Ⅰ forest Ⅱ forest Ⅲ 61% or more forest Ⅱ forest Ⅲ forest Ⅲ

In Table 6 above, the ground is the roughness of the general terrain and the steepness increases from forest I to forest III as the slope and tree density increase. In S802, each agent E (U, G, I) determines a roughness level according to Table 6 above. The level of roughness affects each of the agent E (U, G, I) movement speed and detection range. An example of an influence factor is summarized in Table 7 below.

Humjundo

(이동)

(move)

(Detection range)
ground 1.0 1.0 forest Ⅰ 0.8 0.8 forest Ⅱ 0.6 0.6 forest Ⅲ 0.2 0.6

,

)가 각각 1이므로 이동 속력과 탐지 범위에는 영향이 없다. 험준도가 높아짐에 따라 영향 인자(

,

)는 낮아지므로 각 에이전트 E의 이동 속력과 탐지 범위는 감소하게 된다. 따라서 각 에이전트 E(U, G, I)의 이동 속력은 각각의 평균 기동 속력(

)에 험준도에 따른 이동 영향 인자

을 곱한 값이다.각 에이전트 E(U, G, I)의 감쇠된 탐지 범위는 각각의 최대 탐지 거리(

)와 탐색 영향 인자

를 이용하여 정해지는 범위이다. 탐지 범위와 관련하여 자세한 사항은 도 9에서 후술한다.Referring to Table 7 above, when the roughness is ground, the influence factor (

,

) Is 1, so movement speed and detection range are not affected. Influence factors (

,

) Decreases, so the moving speed and detection range of each agent E decrease. Therefore, the moving speed of each agent E (U, G, I) is the average starting speed of each (

) Influencing factors according to the roughness

The attenuated detection range of each agent E(U, G, I) is the maximum detection distance of each (

) And search influencing factors

It is a range determined using. Details regarding the detection range will be described later in FIG. 9.

를 결정하고 S804에서 다음 위치(

)로 가기 위한 현재 속력 V를

으로 업데이트 한다. S805에서 각 에이전트 E(U, G, I)는 기 업데이트 된 속력에 따라 다음 위치(

)로 이동한다. S806에서 각 에이전트 E(U, G, I)는 패치

의 방문 여부(visitState)를 true로 변경하고 방문 순서(visitSequence)는 이전 패치

의 값보다 1 증가시킨 값으로 업데이트 한다.In S803, each agent E (U, G, I) is an influence factor of the moving speed according to Table 6 above.

The next location in S804 (

) To go to the current speed V

Update. In S805, each agent E(U, G, I) is next position according to the updated speed (

). In S806, each agent E (U, G, I) is patched

Change the visit status (visitState) to true and the visit sequence (visitSequence) is the previous patch.

It is updated with the value increased by 1 from the value of.

While performing steps S801 to S806, in parallel, each agent E(U, G, I) updates the state variable related to the search time.

부터 패치

까지의 이동 시간(

)을 계산한다. 상기 이동 시간(

)은 현재 위치(

)와 다음 위치(

)간의 거리 D를 현재 이동 속력인 V(k+1)로 나눈 값이다.In S810, each agent E (U, G, I) is patched

Patch from

Travel time to

). The travel time (

) Is the current location (

) And next position (

This is the distance between) divided by the current moving speed V(k+1).

)을 업데이트하고 상기 해당 수색 인원 에이전트(I)를 수송한 유인 차량 에이전트(G)는 총 정차 시간(

)을 업데이트 한다(S812).It is checked whether the agent E moving in S811 is one of an unmanned vehicle, a manned vehicle, and a search personnel agent. If the moved agent E is the search personnel agent (I), the search personnel agent (I) is the total search time (

), and the manned vehicle agent (G) who transported the relevant search personnel agent (I) has the total stop time (

) Is updated (S812 ).

또는

)을 업데이트 하고, 이동한 에이전트 E가 유인 차량 에이전트(G)인 경우 S814에서 유인 차량 에이전트(G)는 총 탑승 시간(

)을 추가적으로 업데이트 한다.If the moved agent E is not the search personnel agent (I), in S813, each agent E (U or G) has the total search time (

or

), and if the moved agent E is the manned vehicle agent (G), in S814 the manned vehicle agent (G) is the total boarding time (

).

The total search time (RT) of each agent E (U, G, I) is summarized in Equation 5 below.

까지의 총 수색 시간

에 패치

에서

까지 이동하는 데 소요된 시간인 이동 시간

를 더하여 총 수색 시간

를 업데이트 한다. Unmanned Vehicle Agent (U) Patch Previous

Total search time to

Patch on

in

Travel time, which is the time it took to travel to

Plus the total search time

Update it.

외에 추가로 총 정차 시간

을 더하여 총 수색 시간

를 업데이트 한다. Travel time for manned vehicle agent (G)

In addition, the total stop time

Plus the total search time

Update it.

외에 추가로 총 탑승 시간

을 더하여 총 수색 시간

를 업데이트 한다. Travel Time for Search Personnel Agents (I)

In addition, total boarding time

Plus the total search time

Update it.

9A and 9B are diagrams for specifically explaining the discovery of each agent according to the action rule.

에서 기 설정된 중심각

를 이용하여 부채꼴(circular sector) 넓이 내에 존재하는 패치들을 탐색한다.9A is a flowchart illustrating a process of searching for each agent according to an action rule. In S900, each agent E (U, G, I) is in the current position

Center angle

Use to search for patches that exist within the area of a circular sector.

)에 포함시킨다. In S902, if there is no tree topography in the range, the entire current sector width becomes the detection range. In S907, each agent E(U, G, I) detects the patches included in the current sector width range (

).

)이 360도인 경우 탐색을 마치고 탐지 범위(

)에 포함된 패치들의 탐색 여부(ReconState)를 true로 변경한다. Center angle at S908(

) Is 360 degrees, the search is finished and the detection range (

Changes whether to search for patches included in (ReconState) to true.

)이 360도가 아닌 경우 S909에서 각 에이전트 E(U, G, I)는 상기 중심각의 초기값(theta)만큼 방향(

)을 변경하고 theta만큼 증가된 중심각

을 이용하여 S900 단계부터 이후의 단계를 수행한다. 모든 에이전트(U, G, I)의 가시 범위(FOV)가 360도인 점을 고려하여, 중심각이 360도보다 작을 경우에는 해당 에이전트가 현재 향하고 있는 방향을 theta만큼 이동시키면서 360도 모든 방향에 대해 탐색을 하기 위함이다.Center angle(

) Is not 360 degrees, in S909, each agent E (U, G, I) is directed by the initial value (theta) of the central angle (

) And the center angle increased by theta

The following steps are performed from step S900. Considering that the visible range (FOV) of all agents (U, G, and I) is 360 degrees, if the center angle is less than 360 degrees, the agent will search for all directions in 360 degrees while moving the direction currently facing by theta Is to do.

를 반영하여 탐지범위를 감쇠한다(S903-S907).In S902, if a tree topography exists in the range, each agent E(U, G, I) is a search influencer

Attenuates the detection range by reflecting (S903-S907).

)에서 가장 인접한 수목 지형의 패치(

)를 찾는다. In S903, each agent E (U, G, I) is in the current position (

) The patch of the nearest tree top (

).

)와 상기 수목 지형의 패치(

)간 거리 D를 계산하고, 계산된 거리 D를 이용하여 험준도를 결정한다(S905). In S904, each agent E (U, G, I) is the current location (

) And the patch of the tree topography (

The distance D between) is calculated, and the roughness is determined using the calculated distance D (S905).

를 결정한다. S907에서 각 에이전트 E(U, G, I)는 기 탐색된 패치들 중에서 상기 탐색 영향 인자

를 고려하여 감쇠된 범위에 포함되는 패치들을 선택하여 탐지 범위(

)에 추가한다. 이후 S908 단계를 동일하게 수행하여 방향을 변경하여 추가적으로 탐색하거나 탐색을 종료한다.In S906, each agent E (U, G, I) is a search influence factor according to the determined roughness

Decide. In S907, each agent E (U, G, I) is the search influence factor among the previously discovered patches.

Selecting the patches included in the attenuated range in consideration of the detection range (

). Subsequently, in step S908, the direction is changed to additionally search or end the search.

9B is a diagram for explaining a detection range attenuated according to terrain when each agent searches according to an action rule.

에서 가장 인접한 수목 지형의 패치(

)까지의 거리이고 B는 각 에이전트 E(U, G, I)의 최대 탐지 거리(

)에서 A를 뺀 값이다.Referring to 9b, A is the current position

The patch of the nearest tree terrain in

) And B is the maximum detection distance of each agent E (U, G, I)

) Minus A.

)는 기 설정된 중심각(

)과 최대 탐지 거리(

)에 의한 부채꼴 영역인 반면, 감쇠된 후의 탐지 범위(

)는 상기 중심각(

)과

에 따른 부채꼴 영역이다.Detection range before attenuation (

) Is a preset center angle (

) And maximum detection distance (

), while the detection range after attenuation (

) Is the center angle (

)and

It is a sector of the sector.

As described above, the search behavior modeling apparatus 100 according to the present invention simulates the search behavior with the attenuated detection range according to the characteristics of the terrain, so that a more appropriate search plan can be established under given conditions.

10 is a flowchart for specifically explaining the selection of an unattended vehicle agent's drop-off point according to an action rule.

The unmanned vehicle agent (U) selects an appropriate drop-off point to request the search personnel agent (I) to search for a suspected area (threatening terrain) inside the tree that he or she cannot cover. The unmanned vehicle agent U searches for a suspected area (threatening terrain) that has not yet been searched within a predetermined unit radius while moving along the driving search path.

의 circular sector)에 포함되는 패치들을 인식한다. Referring to FIG. 10, in S1000, the unmanned vehicle agent U moves along the driving search path, and a circle due to a predetermined radius (radius r) around itself (

Of circular sectors).

In S1001, the unmanned vehicle agent U checks whether a patch having a type of terrain is a threat among the recognized patches.

When a patch corresponding to the suspected area exists in S1002, the unmanned vehicle agent U obtains the corresponding patches (threat patches) and the ID (threatAreaID) of the suspected area to which the corresponding patches (threat patches) belong. After (S1013), it is checked whether to search for the suspected area (threatAreaID) (clearState) (S1014).

)를 하차 지점 후보지(

)에 추가하고(S1016) 다음 위치(

)로 이동한 후 S1000으로 돌아가서 이후 단계를 다시 수행한다.In S1015, the unmanned vehicle agent U has not yet been discovered despite the unmanned vehicle agent U discovery (i.e., the patch of the suspected region is searched if the suspected region (threatAreaID) has been cleared (clearState) is false) Not applicable), current location (

Get off at)

) (S1016) and next location (

) And return to S1000 to perform the subsequent steps again.

In step S1015, the unmanned vehicle agent U may end the selection of the disembarkation point because there is no undiscoverable area when the suspected area (threatAreaID) is cleared (clearState) is true.

)의 수를 체크한다(S1003). S1004에서 하차 지점 후보지(

)가 존재하지 않는 경우 하차 지점 선정을 종료할 수 있다. 하차 지점 후보지(

)가 하나 이상 존재하는 경우 무인 차량 에이전트(U)는 하차 지점 후보지(

) 중에서 상기 의심 구역의 패치(threatAreaID)와 가장 인접한 하차 지점 후보지(

)를 결정한다(S1005).In the case where a patch corresponding to the suspected zone does not exist in S1002, the unmanned vehicle agent U is a candidate for a disembarkation point (

) Is checked (S1003). Get off at S1004

) Does not exist, you can end the selection of the drop-off point. Get off point candidate (

If there is more than one, the unmanned vehicle agent (U) is a candidate

), the candidate for the nearest drop-off point to the patch (threatAreaID) in the suspect area (

) Is determined (S1005).

의 하자 지점 여부와 관련된 상태 변수(dismountHere)를 true로 변경함으로써

를 하차지점으로 선정한다(S1006).Unmanned Vehicle Agent (U)

By changing the state variable (dismountHere) related to the defect point of

Is selected as a drop-off point (S1006).

As described above, the unmanned vehicle agent U seeks search cooperation of the search personnel agent I by selecting a disembarkation point for a suspected area outside the scope of his search, and uses a unmanned agent who cooperates with each other. An appropriate search action plan can be established under the given conditions.

11 is a view for explaining a simulation result of a search activity.

According to an embodiment, the search plan storage unit 140 may display real-time search status as shown in FIG. 11. Referring to FIG. 11, the unmanned vehicle agent U selects three drop-off points until reaching the end point, and the manned vehicle agent G stops at the second drop-off point and the search personnel agent I searches It is waiting for a while.

12 is a flowchart illustrating a search behavior modeling method according to another embodiment of the present invention.

The search behavior modeling method according to another embodiment of the present invention simulates a search behavior according to a preset search scenario. Specifically, referring to FIG. 12, the search behavior modeling method includes an agent creation step (S1200), a driving route selection step (S1300), a drop-off point selection step (S1400), a search personnel agent drop-off step (S1500), and an unsearchable area. It may include a search step (S1600) and a search action simulation result storage step (S1700).

In S1200, an environment agent H, an unmanned vehicle agent U, a manned vehicle agent G, and a search personnel agent I are generated.

Specifically, the agent generation unit 110 models the terrain to be searched according to the search scenario to generate an environment agent H. The environmental agent H provides a virtual environment for simulating search activities.

Next, the agent generator 110 models and generates an object performing a search action according to a predetermined behavior rule as a presence or absence agent. The unmanned agent includes an unmanned vehicle agent U, a manned vehicle agent G, and a search personnel agent I performing a search operation according to a predetermined behavior rule in the virtual environment.

The number of the remaining agents U, G, and I other than the environment agent H may be two or more, and may be respectively generated in an appropriate number in consideration of the search target according to the search scenario.

The preset behavior rules are the basis for the behavior of each agent (U, G, I) and a series of simulations for behavior and state changes according to the conditions of the virtual environment recognized by each agent (U, G, I). Is the rule. The preset action rule may be a rule related to at least one action among movement, search, route selection, and drop-off point selection, which are actions for the search operation.

In S1300, the driverless vehicle agent U selects a driving route and searches while moving along the driving route.

Specifically, the unmanned vehicle agent U uses the road network information received from the mock unit 130 to select a driving route according to the preset behavior rule. The road network information is an abstract representation of the main maneuvering path of the unmanned vehicle agent U, and includes information of a node regarding a start position, an end position, and an intersection position.

In the road network information, when there is a path that can be moved between each node, a link is created, and the link has two cost attributes according to the search area and the travel time. The search area cost represents the searchable area when moving to the link. , Search time cost represents the time required to move to the node opposite to the selected link.

The unmanned vehicle agent U moves by selecting an intersection (node) having a maximum search area or a minimum search time as an intermediate destination in consideration of the cost attribute of the link and the search target of the search scenario. Thereafter, when the intermediate vehicle destination U reaches the intermediate destination, the next intermediate destination is again selected in the same manner and moved to the end position (node) to search. Therefore, the driving route is information on the route to the intermediate destination selected by the driverless vehicle agent U.

를 이용하여 자신의 평균 기동 속력을 보정한 후 이동한다. 상기 험준도는 현재 위치와 다음 위치간의 수목 밀도 및 경사도에 따라서 결정될 수 있다.When moving along the driving route, the unmanned vehicle agent U moves after correcting its average starting speed according to a predetermined behavior rule. Specifically, the driverless vehicle agent U determines the roughness level of the terrain between its current location and the next location, and the movement influence factor according to the determined roughness level

Use after correcting his average maneuvering speed. The steepness may be determined according to the tree density and slope between the current location and the next location.

를 이용하여 기 설정된 자신의 탐지 범위(

)를 보정한 후 탐색한다. The driverless vehicle agent U searches according to a predetermined behavior rule when moving along the driving route. Specifically, the driverless vehicle agent (U) has a search effect factor depending on whether or not there is a tree topography at his current location.

Use your preset detection range (

) And search.

The movement and search according to the above-described predetermined action rules are equally applied to the movement and search of the manned vehicle agent U and the search personnel agent I described later.

In S1400, when the unmanned vehicle agent U finds an unsearchable area while moving, a drop-off point is selected on the driving path, and the driving path and the drop-off point are transmitted to the manned vehicle agent.

Specifically, the driving route selected by the unmanned vehicle agent U is transmitted to the manned vehicle agent G. Each agent (U, G, I) transfers information to each other and changes in state due to recognition and behavior of the virtual environment are made through the simulation unit 130.

The unmanned vehicle agent U selects a drop-off point according to a preset behavior rule. When the unmanned vehicle agent U finds an unsearchable area outside the scope of its search, it moves while adding the current location as a drop-off point candidate.

The non-searchable area is a suspected area (threatening terrain) previously set according to the search scenario, and means an area that has not yet been searched despite the search of the unmanned vehicle agent U.

The unmanned vehicle agent U selects the candidate point of disembarkation point closest to the non-searchable area from among the previously added drop-off point candidates and delivers it to the manned vehicle agent G.

In S1500, after the manned vehicle agent G receives the driving route and the disembarking point, it moves along the driving route to disembark the search personnel agent I to the disembarking point. Specifically, the manned vehicle agent G moves to the disembarkation point received from the unmanned vehicle agent U according to a predetermined behavior rule. At the alighting point, the manned vehicle agent G stops and the search personnel agent I gets off. The manned vehicle agent G waits for the search personnel agent I to board again.

In S1600, the search personnel agent I searches for the unsearchable area and boards the manned vehicle agent. Specifically, the search agent agent (I) moves to the unsearchable area at an average maneuvering speed corrected according to a predetermined behavioral rule at the drop-off point. The search personnel agent (I) searches according to its detection range corrected according to a predetermined behavior rule. After completing the search, the search personnel agent (I) moves to the manned vehicle agent (G) that is stopping and boards.

Specifically, the S1300 to S1600 are repeated until the driverless vehicle agent U reaches the end position and all suspected areas (threatening terrain) are searched.

In S1700, simulation results of the search activity are stored. Specifically, the search plan storage unit 140 stores simulation results of search actions performed according to the S1300 to S1600.

The search method storage unit 140 receives and stores the state variables of all agents H, U, G, and I each time an action for a search operation of each agent U, G, I is finished. Therefore, the simulation result of the search activity includes information of the grid map and the state variable of the environment agent H, and at least one of the state variables of the unmanned vehicle agent U, manned vehicle agent G, and search personnel agent I The state variable of can be included.

According to an embodiment, the simulation result of the search activity includes information related to the environmental agent H and at least one of the total search area, total search time, and movement path information of each agent U, G, and I. It may include.

According to an embodiment, the search behavior modeling method according to the present invention may further include an analysis step of analyzing a simulation result of the search behavior according to a preset criterion.

In the analysis step, the search plan analysis unit 150 analyzes the simulation result of the search action according to a preset criterion. The preset criteria may include at least one of search efficiency, search time required, and contribution of each agent (U, G, I) to color behavior.

As described above, the search behavior modeling method according to another embodiment of the present invention can consistently simulate and analyze search behaviors using agents that perform detailed search behaviors according to predetermined behavior rules, without analyst intervention. You can get more objective simulation results.

In addition, since the agent performing the search operation according to the preset behavioral rule of the present invention has different movement speed and detection range according to the characteristics of the terrain, the accuracy of the simulation result is improved.

In addition, the agent (U) performing a search operation according to the preset behavioral rules of the present invention is selected as a drop-off point for an area where the search is impossible, and the other agent (I) performs a search, so that the presence or absence of agents cooperating with each other It can be used not only to establish an appropriate search activity plan under the given conditions, but also to be used for other purposes, such as mock disaster and crime response plans.

In addition, the present invention analyzes a simulation result of a search activity according to a preset criterion, and a more appropriate search plan can be established by comparing and analyzing a plurality of simulation results.

The above detailed description should not be construed as limiting in all respects, but should be considered illustrative. The scope of the invention should be determined by rational interpretation of the appended claims, and all changes within the equivalent scope of the invention are included in the scope of the invention.

100: search behavior modeling device
110: agent generation unit
120: control unit
130: mock
140: search room storage
150: search plan analysis unit

Claims (10)

In the search environment modeling method to simulate the search behavior according to a predetermined search scenario in a virtual environment,
A first step of generating an environment agent providing the virtual environment, an unmanned vehicle agent performing a search operation in the virtual environment, a manned vehicle agent, and a search personnel agent;
A second step in which the driverless vehicle agent selects a driving route and searches while moving along the driving route;
A third step of selecting a disembarkation point on the driving route and transferring the driving route and the disembarking point to the manned vehicle agent when the unmanned vehicle agent finds an unsearchable area while moving;
A fourth step of, after the manned vehicle agent receives the driving route and the disembarking point, moves along the driving route and unloads the search personnel agent at the disembarking point;
A fifth step in which the search agent agent boards the manned vehicle agent after searching the unsearchable area; And
And a sixth step of storing a simulation result of the search activity according to the second to fifth steps,
The second to fifth steps are search behavior modeling method characterized in that it is performed according to a predetermined behavior rule. According to claim 1,
The preset action rule is a search action modeling method characterized in that the action for the search operation is a rule related to at least one action among movement, search, route selection, and drop-off point selection. According to claim 2,
In the second to fifth steps,
The unmanned vehicle, manned vehicle and search personnel agent,
Each preset average maneuvering speed
Searching behavior modeling method, characterized in that, when moving according to the preset behavior rule, the average maneuvering speed is corrected and moved according to the terrain roughness between the current location and the next location. According to claim 2,
In the second and fifth steps,
The driverless vehicle and search personnel agent,
Each preset detection range has
Searching behavior modeling method, characterized in that, when searching at one's current location according to the preset behavior rule, each of the preset detection range is corrected and then searched according to the presence or absence of tree topography at the current location. According to claim 2,
In the second step,
The unmanned vehicle agent selects a driving route having a maximum search area or a minimum search time according to the preset behavior rule. According to claim 2,
The third step,
The unmanned vehicle agent adding a disembarkation point candidate to move while adding the current location to the disembarkation point candidate when the unsearchable area is found while moving along the driving path; And
And selecting a disembarkation point to select a disembarkation point candidate closest to the unsearchable area as the disembarkation point among the disembarkation point candidates previously added by the driverless vehicle agent. According to claim 1,
The simulation result of the search activity,
Characterized in that it includes at least one of the total search area searched by each of the unmanned vehicle, manned vehicle, and search personnel agent, the total search time required for each search, information on each movement path, and information related to the environment agent. Search behavior modeling method. The method of claim 7,
Further comprising a sixth step of analyzing the simulation results of the search activity according to a predetermined standard
The preset criterion includes a search behavior modeling method, characterized in that it includes at least one of the contribution efficiency of each of the unmanned vehicle, manned vehicle, and search personnel agent to search efficiency, search time, and search activity. In the search behavior modeling device that simulates a search behavior according to a preset search scenario in a virtual environment,
An agent generating unit that generates an environment agent providing the virtual environment, an unmanned vehicle agent performing a search operation in the virtual environment, a manned vehicle agent, and a search personnel agent;
A control unit for controlling an unmanned vehicle agent, a manned vehicle agent, and a search personnel agent to act according to a predetermined behavior rule in the virtual environment;
A simulation unit that simulates the search behavior by providing information related to the environment agent and at least one of an unmanned vehicle agent, a manned vehicle agent, and a search personnel agent to an unmanned vehicle agent, a manned vehicle agent, and a search personnel agent, respectively; And
And a search room storage unit for storing a simulation result of the search activity,
The driverless vehicle agent selects a driving route, searches while moving along the driving route, selects a drop-off point on the driving route when it finds an unsearchable area while moving, and determines the driving route and the drop-off point. To the mock department,
The manned vehicle agent receives the driving route and the getting-off point from the simulation unit, moves along the driving route, and unloads the search personnel agent at the getting-off point,
And the search agent agent searches for the unsearchable area and then boards the manned vehicle agent. The method of claim 9,
The preset action rule is a search action modeling device, characterized in that the action related to at least one action among movement, search, route selection, and drop-off point selection, which are actions for the search operation.

Images