Evacuation route determination method, device, electronic device and storage medium
By using deep reinforcement learning algorithms to update evacuation routes in real time in public places, the problem of evacuation route planning in complex accident environments has been solved, achieving safe and efficient evacuation route planning and reducing casualties.
Patent Information
- Application Number
- CN202210112873.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-01-29
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2042-01-29
AI Technical Summary
In complex public places, evacuation route planning is difficult to implement effectively in the event of accidents such as fires and collapses, leading to an increase in casualties.
By constructing an evacuation route determination method, an accident map is generated using building layout information and accident information. Combined with deep reinforcement learning algorithms such as cooperative dual deep Q-network, the evacuation routes are updated in real time. Taking into account target distribution, environmental changes and accident parameters, the optimal evacuation route is planned.
It improved the safety and real-time performance of evacuation routes, reduced casualties, and decreased accident losses.
Smart Images

Figure CN114510545B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of information technology, and in particular to an evacuation path determination method, device, electronic device and storage medium. Background Art
[0002] As the internal structures of public places become increasingly complex and the scale of buildings becomes larger, the safety of public places has become an issue that cannot be ignored. Accidents such as fire and collapse are one of the major disasters that threaten the safety of public places. These accidents can easily lead to the safety of property and personnel. Therefore, timely evacuation of people in buildings can at least reduce casualties. Summary of the invention
[0003] In view of this, embodiments of the present application hope to provide an evacuation path determination method, device, electronic device and storage medium.
[0004] The technical solution of this application is implemented as follows:
[0005] In a first aspect, an embodiment of the present application provides a method for determining an evacuation path, comprising:
[0006] Inputting the first model with the building layout information and the accident information at the accident scene to generate the accident map information at the current moment; wherein the accident map information at least indicates whether different areas of the accident scene are accessible;
[0007] Obtaining distribution information of targets at the accident scene and exit information of emergency exits;
[0008] Determining relative position information between the target and the safety exit according to the distribution status information;
[0009] A safe evacuation path for the target to leave the accident scene is obtained according to the accident map information, the relative position information and the second model.
[0010] Based on the above solution, the method further includes:
[0011] After the movement status of the target along the safe evacuation path satisfies the first condition, the distribution status information of the target at the accident scene is reacquired.
[0012] Based on the above solution, after the movement status of the target along the safe evacuation path meets the first condition, re-obtaining the distribution status information of the target at the accident scene includes:
[0013] After the movement status of the target according to the safe evacuation path satisfies the first condition, determining whether the relative position relationship between the target and the safe exit satisfies a second condition;
[0014] If the relative position relationship between the target and the safety exit does not satisfy the second condition, the distribution status information of the target at the accident scene is re-acquired.
[0015] Based on the above solution, the method further includes:
[0016] If the relative position between the target and the safety exit satisfies the second condition, the determination of the safe evacuation path is stopped.
[0017] Based on the above solution, obtaining a safe evacuation path for the target to leave the accident scene according to the accident map information, the relative position information and the second model includes:
[0018] Determining environmental information of the area where the target is located from the accident map information according to the distribution status information;
[0019] The environmental information and the relative position information are input into the second model to obtain a safe evacuation path for the target to leave the accident scene output by the second model.
[0020] Based on the above solution, the step of inputting the environmental information and the relative position information into the second model to obtain the safe evacuation path for the target to leave the accident site output by the second model includes:
[0021] Determining a plurality of alternative movement paths that meet safety conditions according to the accident map information and the relative position information;
[0022] When there are multiple alternative movement paths, the alternative movement path with the greatest safety benefit is selected as the safe evacuation path; wherein the safety benefit has one of the following specific characteristics:
[0023] The safety benefit is negatively correlated with the distance the target moves to the safety exit along the corresponding alternative moving path;
[0024] The safety benefit is negatively correlated with the accident rate predicted when the target moves to the safety exit according to the corresponding alternative moving path;
[0025] The safety benefit is positively correlated with the probability that the target successfully reaches the safety exit along the corresponding alternative moving path.
[0026] Based on the above solution, the target distribution information indicates:
[0027] The distribution positions of the targets and / or the number of targets located within the accident scene;
[0028] The environmental information at least indicates the area type of the adjacent area where the target is located;
[0029] The area types include: a passable area and a non-passable area outside the passable area.
[0030] Based on the above solution, the non-traffic area includes at least one of the following:
[0031] Obstacle area, an area where obstacles that hinder passage are piled up;
[0032] Accident area, the area where the accident occurred;
[0033] A congested area is an area that accommodates more than a preset number of targets.
[0034] Based on the above solution, the method further includes:
[0035] Acquiring accident parameters of the accident scene;
[0036] According to the accident parameters, it is determined whether an adjacent area of the target location is the accident area.
[0037] Based on the above scheme, the accident parameter includes at least one of the following:
[0038] Fire parameters;
[0039] Collapse parameters;
[0040] Flood parameters.
[0041] Based on the above scheme, the accident scene is a fire scene;
[0042] The obtaining of accident parameters at the accident scene includes:
[0043] The type of fire source at the fire scene, the spatial parameters of the space where the fire source is located, and the heat release rate of the fire source are obtained, wherein the spatial parameters of the space where the fire source is located indicate the ignition point position of the fire source and / or the distribution status of combustibles in the space where the fire source is located.
[0044] Based on the above solution, determining whether the adjacent area where the target is located is the accident area according to the accident parameters includes:
[0045] Input a third model according to the type of the fire source at the fire scene, the spatial parameters of the space where the fire source is located, and the heat release rate of the fire source to obtain an output result;
[0046] According to the output result, obtaining the temperature and / or visibility of each area of the fire scene;
[0047] Determining a dangerous area of the fire scene according to the temperature and / or visibility;
[0048] According to the dangerous area, it is determined whether an adjacent area where the target is located is the accident area.
[0049] Based on the above solution, determining the dangerous area of the fire scene according to the temperature and / or visibility includes:
[0050] When the temperature exceeds a first threshold, and / or when the visibility is lower than a second threshold, the adjacent area where the target is located is determined to be a dangerous area of the fire scene.
[0051] In a second aspect, an embodiment of the present application provides an evacuation path determination device, including:
[0052] A generating module, configured to input the first model according to the building layout information and the accident information of the accident scene, and generate the accident map information at the current moment; wherein the accident map information at least indicates whether different areas of the accident scene are passable;
[0053] A first acquisition module is used to acquire the distribution status information of the accident scene targets and the exit information of the safety exit;
[0054] A first determining module, configured to determine relative position information between the target and the safety exit according to the distribution status information;
[0055] The obtaining module is used to obtain a safe evacuation path for the target to leave the accident scene according to the accident map information, the relative position information and the second model.
[0056] In a third aspect, an embodiment of the present application provides an electronic device, including:
[0057] a memory storing computer-readable instructions;
[0058] A processor, connected to the memory, is used to implement the method provided by any technical solution of the first aspect by executing computer executable instructions stored in the memory.
[0059] In a fourth aspect, an embodiment of the present application provides a computer storage medium, wherein the computer storage medium stores computer executable instructions, and after the computer executable instructions are executed, the method provided by any technical solution of the first aspect can be implemented.
[0060] The evacuation path determination method, device, electronic device and storage medium provided in the embodiment of the present application can generate the accident map information at the current moment according to the input of the building layout information and accident information of the accident scene into the first model; wherein the accident map information at least indicates whether different areas of the accident scene are accessible; obtains the distribution status information of the target at the accident scene and the exit information of the safety exit; determines the relative position information between the target and the safety exit according to the distribution status information; obtains the safe evacuation path for the target to leave the accident scene according to the accident map information, the relative position information and the second model. In this way, according to the building layout information and the accident information of the accident scene, the disaster situation of different areas in the entire building can be known through the first model, so that the optimal evacuation path can be given in combination with the location of the target and the location of the safety exit, etc. Compared with the target judging the disaster situation based on its own line of sight, it can reduce the casualties caused by going to the area with more serious accidents, and reduce the losses caused by the accident. BRIEF DESCRIPTION OF THE DRAWINGS
[0061] Figure 1 A schematic diagram of a flow chart of a method for determining an evacuation path provided in an embodiment of the present application;
[0062] Figure 2 A flowchart of another method for determining an evacuation path provided in an embodiment of the present application;
[0063] Figure 3 A flowchart of a fire evacuation simulation method based on deep reinforcement learning provided in an embodiment of the present application;
[0064] Figure 4 A schematic diagram of a fire scene model established using fire dynamics simulation software provided in an embodiment of the present application;
[0065] Figure 5 A schematic diagram of a grid simulation model for an evacuation scenario provided in an embodiment of the present application;
[0066] Figure 6 A schematic diagram of personnel evacuation provided in an embodiment of the present application;
[0067] Figure 7 A schematic diagram of the end of personnel evacuation provided in an embodiment of the present application;
[0068] Figure 8 A schematic diagram of the structure of an evacuation path determination device provided in an embodiment of the present application;
[0069] Fig. 9 It is a structural schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0070] In order to enable a more detailed understanding of the features and technical contents of the present application, the implementation of the present application is described in detail below in conjunction with the accompanying drawings. The attached drawings are for reference only and are not used to limit the present application.
[0071] like Figure 1 As shown, an embodiment of the present application provides a method for determining an evacuation path, comprising:
[0072] Step S110: inputting the first model according to the building layout information and the accident information of the accident scene to generate the accident map information at the current moment; wherein the accident map information at least indicates whether different areas of the accident scene are passable;
[0073] Step S120: Obtaining the distribution information of the accident scene targets and the exit information of the emergency exits;
[0074] Step S130: determining the relative position information between the target and the safety exit according to the distribution status information;
[0075] Step S140: obtaining a safe evacuation path for the target to leave the accident scene according to the accident map information, the relative position information and the second model.
[0076] The accident scene may be a space where the accident occurs.
[0077] Exemplarily, the accident scene includes but is not limited to:
[0078] The fire scene where the fire occurred;
[0079] The earthquake site where the earthquake occurred;
[0080] the collapse site where the collapse occurred;
[0081] A flood scene with abnormal water flow.
[0082] The building layout information may indicate the overall layout of the building, and may specifically indicate information such as the location, orientation, and surrounding environment of all facilities in the building.
[0083] Exemplarily, the building layout information may indicate the location of a safety passage in the building, and / or indicate a location blocked by a wall in the building.
[0084] In the embodiment of the present application, the building layout information includes, but is not limited to: building design plans, completion acceptance drawings or data obtained from on-site measurements, and other information that can obtain the overall layout of the building.
[0085] The accident information may include any information describing the accident, for example, information indicating the process and conditions of the occurrence and development of the accident. For example, taking a fire scene as an example, the accident information may include information such as the location of the fire source, the type of fire source, the severity of the fire, and the area covered by the fire.
[0086] The first model may be a simulation model of an evacuation scenario, which is used to generate overall accident map information based on input information. The first model may grid the entire accident scene based on building layout information, that is, divide the space where the accident scene is located into many areas, and then determine the specific conditions of each area based on the input accident information, for example, divide each area into a passable area and a non-passable area.
[0087] The communication area may be an area through which the target can pass safely;
[0088] The non-communication area may be an area where the target cannot pass or where a safety accident is likely to occur when passing through.
[0089] The accident map information is used to indicate the specific conditions of different areas of the accident scene, such as passable areas and non-passable areas, etc. The passable area may be an area where the target can pass normally; the non-passable area may be an area where the target cannot pass normally, such as an area blocked by obstacles, or a dangerous area caused by an accident, or an area that has reached the limit of the number of targets.
[0090] The target may include: various movable objects, and illustratively, the target may include: living things, such as people and / or pets. In another illustrative example, the target may also include: ground mobile robots. Since the ground mobile robots themselves are mobile, leaving in time when an accident occurs can reduce the damage to the ground mobile robots.
[0091] The target distribution status information is used to indicate the target distribution status, and the distribution status may include: distribution position and / or distribution density, etc.
[0092] The distribution status of the targets may be the number of the targets and / or the locations of the targets in the accident scene.
[0093] The exit information of the emergency exit is used to indicate the location of the emergency exit. For example, the emergency exit may include: an elevator emergency exit, a staircase emergency exit, or an underground passage emergency exit of a building.
[0094] The safety exit may be an evacuation staircase that complies with regulations or an exit for safely escaping from the accident scene. In an embodiment of the present application, the target may safely escape from the accident scene through the safety exit.
[0095] The relative position information is used to indicate the relative position between the target and the safety exit.
[0096] The second model can specifically be a neural network model, which is used to calculate the safe evacuation path from the target to the safe exit based on the input information. In an embodiment of the present application, a collaborative dual-depth Q network algorithm can be used as a neural network model for deep reinforcement learning. The collaborative dual-depth Q network algorithm is formed by optimizing the neural network input layer and sharing multi-objective neural networks and experience on the basis of a traditional independent deep Q network algorithm, thereby improving the stability of the algorithm.
[0097] The safe evacuation path may be a moving path of the target through the passage area to the safe exit. In an embodiment of the present application, the safe evacuation path includes a safe evacuation path for the target as a whole and a safe evacuation path corresponding to each target.
[0098] In an embodiment of the present application, the method further includes:
[0099] After the movement status of the target along the safe evacuation path satisfies the first condition, the distribution status information of the target at the accident scene is reacquired.
[0100] The movement status is used to indicate the movement of the target, that is, to indicate the change of the position of the target; the movement can be the behavior of the target changing its original position, such as walking or running or moving by indoor transportation, etc.
[0101] The first condition may be the time or distance of movement of the target. For example, the movement condition satisfying the first condition may include at least one of the following:
[0102] When the target moving steps reach the step threshold, it can be considered that the first condition is met;
[0103] The target's current moving duration reaches the duration threshold, which can be considered to meet the first condition;
[0104] The distance traveled by the target this time reaches the distance length threshold, which can be considered to meet the first condition;
[0105] When the displacement of the target reaches the displacement threshold, it can be considered that the first condition is satisfied.
[0106] In the embodiment of the present application, the number of targets and the number of safety exits can be one or more (two or more), so the nearest safety exit may change at any time during the movement of the target, so the best safe evacuation route needs to be re-determined. In this way, by updating the safe evacuation path of the target in real time according to the time and / or distance of the target movement, the safety and real-time performance are better.
[0107] In the embodiment of the present application, after the movement status of the target according to the safe evacuation path satisfies the first condition, reacquiring the distribution status information of the target at the accident scene includes:
[0108] After the movement status of the target according to the safe evacuation path satisfies the first condition, determining whether the relative position relationship between the target and the safe exit satisfies a second condition;
[0109] If the relative position relationship between the target and the safety exit does not satisfy the second condition, the distribution status information of the target at the accident scene is re-acquired.
[0110] The second condition may be the distance between the target and the safety exit. If the distance between the target and the safety exit satisfies the second condition, it means that the target has reached the safety exit, and there is no need to re-acquire the distribution status information of the target at the accident site.
[0111] In an embodiment of the present application, the method further includes:
[0112] If the relative position between the target and the safety exit satisfies the second condition, the determination of the safe evacuation path is stopped.
[0113] If the distance between the target and the safe exit meets the second condition, it means that the target has reached the safe exit or compared with the last comparison of the distance between the target and the safe exit and can reach the safe exit after passing through one or more determined passage areas, and there is no need to determine the safe evacuation path of the target, thereby saving computing resources and calculating the safe evacuation paths of the remaining targets more quickly.
[0114] like Figure 2 As shown, obtaining a safe evacuation path for the target to leave the accident scene based on the accident map information, the relative position information and the second model includes:
[0115] Step S210: determining the environmental information of the area where the target is located from the accident map information according to the distribution status information;
[0116] Step S220: input the environmental information and the relative position information into the second model to obtain a safe evacuation path for the target to leave the accident scene output by the second model.
[0117] The environmental information is used to at least indicate the area type of the adjacent area where the target is located. During the evacuation of the target, the accident may still be developing, which may cause some original passable areas to become non-passable areas. Therefore, combined with the environmental information of the area where the target is located, the safe evacuation path for the target to leave the accident site can be better obtained. In this way, the corresponding safe evacuation path can be calculated in real time according to the changes in the accident site, which is safer and more practical.
[0118] In an embodiment of the present application, inputting the environmental information and the relative position information into the second model to obtain a safe evacuation path for the target to leave the accident scene output by the second model includes:
[0119] Determining a plurality of alternative movement paths that meet safety conditions according to the accident map information and the relative position information;
[0120] When there are multiple alternative movement paths, the alternative movement path with the greatest safety benefit is selected as the safe evacuation path; wherein the safety benefit has one of the following specific characteristics:
[0121] The safety benefit is negatively correlated with the distance the target moves to the safety exit along the corresponding alternative moving path;
[0122] The safety benefit is negatively correlated with the accident rate predicted when the target moves to the safety exit according to the corresponding alternative moving path;
[0123] The safety benefit is positively correlated with the probability that the target successfully reaches the safety exit along the corresponding alternative moving path.
[0124] The safety condition may be that there is only a passable area on the alternative moving path. If there is a non-passable area, it means that the target cannot move to the safe exit according to the alternative moving path, and the safety condition is not met.
[0125] The alternative moving path may be an area that the target needs to pass through.
[0126] The safety benefit may be the safety benefit obtained by the target moving along the alternative moving path. If the safety benefit increases, it means that the target is safer, that is, the probability of casualties occurring at the accident site is lower.
[0127] The positive correlation may include but is not limited to: positive proportional correlation, etc.
[0128] The negative correlation may include but is not limited to: negative proportional correlation and the like.
[0129] In the embodiment of the present application, the target distribution information indicates: the distribution position of each target and / or the number of targets located in the accident scene;
[0130] The environmental information at least indicates the area type of the adjacent area where the target is located;
[0131] The area types include: a passable area and a non-passable area outside the passable area.
[0132] In the embodiment of the present application, the non-traffic area includes at least one of the following:
[0133] Obstacle area, an area where obstacles that hinder passage are piled up;
[0134] Accident area, the area where the accident occurred;
[0135] A congested area is an area that accommodates more than a preset number of targets.
[0136] In the embodiment of the present application, taking into account the influence of each other among the targets, a congestion area is set in the non-traffic area to limit the number of targets in the area and ensure the smooth escape of the targets, thereby reducing accidents such as congestion and trampling.
[0137] The preset number may be determined based on the area of the region and / or the size of the channel in the region. For example, the area of the region may be positively correlated with the preset number; or, the volume of the channel in the region may be positively correlated with the preset number.
[0138] In an embodiment of the present application, the method further includes:
[0139] Acquiring accident parameters of the accident scene;
[0140] According to the accident parameters, it is determined whether an adjacent area of the target location is the accident area.
[0141] The accident parameter may be one of the aforementioned accident information, and is used to indicate the nature of the accident. The occurrence and possible development process of the accident can be determined through the accident parameter, so that the type information of the region can be updated accordingly.
[0142] For example, various security equipment is installed in the building, and these security equipment can collect information from the accident scene to obtain accident information. For example, the camera in the building can collect images of the accident scene, and through image analysis, the fire or collapse situation at the fire scene and / or collapse scene, as well as the target distribution situation, can be known. In addition, smoke sensors in the building can collect smoke information and report it to the network side. Electronic devices can obtain accident information through the network or extract accident parameters based on the accident information, that is, electronic devices directly obtain accident information from network devices or directly receive accident parameters.
[0143] The accident area may be an area where an accident occurs or is affected by an accident, such as an area where a fire occurs, an area where smoke generated by the fire spreads and affects visibility, an area where an earthquake causes house collapse, etc.
[0144] In the embodiment of the present application, the accident parameter includes at least one of the following:
[0145] Fire parameters;
[0146] Collapse parameters;
[0147] Flood parameters.
[0148] In the embodiment of the present application, the accident scene is a fire scene;
[0149] The obtaining of accident parameters at the accident scene includes:
[0150] The type of fire source at the fire scene, the spatial parameters of the space where the fire source is located, and the heat release rate of the fire source are obtained, wherein the spatial parameters of the space where the fire source is located indicate the ignition point position of the fire source and / or the distribution status of combustibles in the space where the fire source is located.
[0151] The fire scene may be a fire accident scene.
[0152] The fire source may be an energy source that causes combustibles, combustion-supporting materials, and certain explosive substances to burn or explode, such as thermal energy, electrical energy, mechanical energy, chemical energy, light energy, and the like.
[0153] The types of the fire source may be open flame, high temperature object, electric heat energy, chemical heat energy, mechanical heat energy, biological heat, light energy, nuclear energy, etc.
[0154] The heat release rate of the fire source may be the heat released by the combustion of the material of the fire source per unit time, and is used to indicate the propagation speed of the flame generated by the combustion of the fire source.
[0155] The ignition point position of the fire source indicates the initial position where the fire source burns.
[0156] The distribution status of the combustibles in the space where the fire source is located at least indicates the position and quantity of the combustibles.
[0157] The collapse parameters may include the source intensity and source depth of the earthquake source at the collapse site.
[0158] The earthquake source may be a place where the vibration is caused by the rupture of rock layers inside the earth.
[0159] The earthquake source intensity may be the magnitude of earthquake energy, and is used to indicate the severity of an earthquake.
[0160] The earthquake source depth may be the distance from the earthquake source vertically upward to the ground surface, and is used to indicate the location of the earthquake source. Flood parameters may include the flow rate, water level, water volume, etc. of the water source at the flood site.
[0161] The water source may be a source of water that causes a flood.
[0162] The flow rate may include average flow rate and instantaneous flow rate, which are used to indicate the severity and propagation speed of the flood.
[0163] The water level may be the elevation of the free water surface of the water source above a fixed base surface, which can directly reflect the increase or decrease in the amount of water in the water source and is used to indicate the probability of a flood occurring.
[0164] The water volume may be the flow rate of the water source.
[0165] In an embodiment of the present application, determining whether an adjacent area of the target location is the accident area according to the accident parameter includes:
[0166] Input a third model according to the type of the fire source at the fire scene, the spatial parameters of the space where the fire source is located, and the heat release rate of the fire source to obtain an output result;
[0167] According to the output result, obtaining the temperature and / or visibility of each area of the fire scene;
[0168] Determining a dangerous area of the fire scene according to the temperature and / or visibility;
[0169] According to the dangerous area, it is determined whether an adjacent area where the target is located is the accident area.
[0170] The third model may be a fire scene model, which is used to simulate the occurrence and development of a fire according to the input fire source type, the spatial parameters of the space where the fire source is located, and the heat release rate of the fire source, and obtain an output result. Specifically, the fire scene is modeled using fire dynamics simulation software to simulate the occurrence and development of a fire by setting the fire source type and location, heat release rate, and environmental parameters.
[0171] The output result may be a result obtained by the third model processing the input information.
[0172] The temperature may be a physical quantity representing the degree of hotness or coldness, and is used to indicate whether there is a fire in the area.
[0173] The visibility may be a maximum distance representing a target identification object, and is used to indicate whether there is smoke caused by a fire in the area.
[0174] The dangerous area refers to an inaccessible area that is difficult for targets to pass through due to the occurrence and development of fire.
[0175] In an embodiment of the present application, determining the dangerous area of the fire scene according to the temperature and / or visibility includes:
[0176] When the temperature exceeds a first threshold, and / or when the visibility is lower than a second threshold, the adjacent area where the target is located is determined to be a dangerous area of the fire scene.
[0177] In the embodiment of the present application, the temperature and / or visibility are used to determine whether the area is affected by the fire. The first threshold and the second threshold can be set manually as needed. In practical applications, sensors and other devices can be used to detect the temperature and visibility in the area to determine whether it is a dangerous area.
[0178] Based on the above embodiments, the embodiments of the present application provide a fire evacuation simulation method and system based on deep reinforcement learning, which optimizes the neural network input layer and multi-mobile neural network sharing and experience sharing on the basis of the traditional independent deep Q network (IDQN) algorithm, forming a collaborative dual deep Q network algorithm, which improves the stability of the algorithm and enables it to plan the evacuation path of personnel in a dynamically changing fire environment. In the embodiments of the present application, the mobile body can learn environmental data during the interaction with the fire environment without manual input, which increases the safety of the evacuation path.
[0179] The present application is further described below in conjunction with the accompanying drawings and embodiments.
[0180] It should be noted that the following detailed descriptions are illustrative and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.
[0181] like Figure 1 As shown, the embodiment of the present application provides a flow chart of a fire evacuation simulation method based on deep reinforcement learning, including:
[0182] Step S310: Use fire dynamics simulation software to establish a fire scene model, and obtain the real-time danger zone affected by high temperature and fire smoke from the results of the fire numerical simulation, wherein the fire scene model is one of the aforementioned third models, the results of the fire numerical simulation are one of the aforementioned output results, and the real-time danger zone is one of the aforementioned danger zones.
[0183] like Figure 4 As shown, a schematic diagram of a fire scene model established using fire dynamics simulation software provided in an embodiment of the present application is provided. According to the architectural design plan, the completion acceptance drawing and the field measurement data, the fire scene model is established using the fire dynamics simulation software, the fire source type and location, the heat release rate and the environmental parameters are set, and the occurrence and development process of the fire is simulated using the fire dynamics simulation software, where the fire source type and location, the heat release rate and the environmental parameters are one of the aforementioned accident parameters.
[0184] A slice is set at 1.5m from the ground to observe the temperature distribution and visibility distribution of the entire indoor area, and the area with a temperature higher than 65°C or a visibility lower than 8m is set as a fire hazard area. From the beginning of the fire, the distribution of dangerous areas affected by high temperature and fire smoke is obtained every 1s. The 65°C is one of the first thresholds, and the 8m is one of the second thresholds.
[0185] Step S320: Establish an evacuation scenario simulation model based on real-scene information, and set information such as the number of evacuees and the location of each person, wherein the real-scene information is a type of building layout information of the aforementioned accident scene, the evacuation scenario simulation model is a type of the aforementioned first model, the evacuees are a type of the aforementioned targets, and the information such as the number of evacuees and the location of each person is a type of distribution status information of the targets at the accident scene.
[0186] like Figure 5 As shown, a schematic diagram of a grid simulation model of an evacuation scenario provided in an embodiment of the present application is provided. Based on real scene information such as the architectural design plan, a grid simulation model of the evacuation scenario is established with a side length of 0.4 m for the grid unit. The model includes walls, emergency exits, real-time fire danger areas, and evacuees, and the wall is one of the aforementioned obstacles.
[0187] The number and location distribution of evacuees can be set according to the actual situation, and can also be changed as needed. Evacuees should have the following characteristics:
[0188] 1. Personnel need to know the location of the exit, but do not know the distribution of obstacles and the real-time distribution of fire danger areas, which requires learning;
[0189] 2. During the evacuation process, each person occupies the space of a grid, that is, a grid can only accommodate one person at most;
[0190] 3. During the evacuation process, personnel can move one grid in eight directions each time: up, down, left, right, upper left, lower left, upper right, and lower right. They cannot move multiple grids at a time.
[0191] 4. When the next grid after the personnel choose the moving direction is an obstacle grid or has been occupied by other personnel, the evacuated personnel will stay in the original grid, and this action will be regarded as staying in place.
[0192] Step S330: As the evacuation time increases, the real-time fire danger zone is imported into the evacuation model.
[0193] When a fire just occurs, the area of the dangerous area affected by the fire is small, and the impact on the evacuation of personnel is small. As the evacuation time increases, the area of the dangerous area affected by the fire gradually expands, which has a greater impact on the evacuation route selection of personnel who have not yet completed the evacuation. Therefore, the evacuation scene of the evacuation model is updated in real time at a time interval of 1s, thereby realizing the simulation of the real evacuation scene and enhancing the authenticity and safety of the evacuation process. The 1s time interval is one of the aforementioned first conditions.
[0194] The impact of the imported real-time fire danger zone on personnel evacuation is similar to that of obstacles. When the next grid after the personnel chooses the moving direction is the fire danger zone grid, the evacuated personnel will stay in the original grid. However, the fire danger zone is constantly changing. When the personnel evacuation route overlaps with the fire danger zone, it will hinder the evacuation of personnel to the exit, and may even cause the personnel to be surrounded by the fire danger zone and unable to reach the exit. This requires the evacuation algorithm to plan the optimal evacuation path based on full consideration of the real-time changes of the fire danger zone. The optimal evacuation path is one of the aforementioned safe evacuation paths.
[0195] Step S340: Utilize the collaborative dual-depth Q network algorithm to perform optimal path planning for personnel.
[0196] During the evacuation of all moving bodies, all moving bodies that have not reached the exit are screened out based on the distance between each person and the nearest exit. Each moving body collects environmental information and selects an action based on the current training results of the collaborative dual-depth Q network algorithm. However, when the moving body is on the grid of the fire hazard area, the moving body does not select an action and stays in place. The moving body is one of the aforementioned targets. After each moving body selects an action, a reward value is calculated, and the action is executed according to the selection, so that each moving body enters the next state. The execution of the action is one of the aforementioned first conditions. Then, each experience element is stored in the experience pool, and some experience elements in the experience pool are selected for learning and training by random batch sampling, and the neural network parameters θ are updated. Finally, when all personnel have escaped from the scene or the evacuation time step reaches the set upper limit, this round of model learning and training is stopped, and the next round of learning and training is carried out until the set number of training rounds is reached, such as Figure 6 As shown, it is a schematic diagram of personnel evacuation provided in an embodiment of the present application.
[0197] Furthermore, a collaborative dual-depth Q-network algorithm is adopted as the neural network model of deep reinforcement learning. The environmental information of the eight neighborhoods around the mobile body is added to the neural network input layer to simulate the observation of the surrounding environment by people in a low-visibility environment, thereby establishing a homogeneous mobile body and optimizing the neural network input layer.
[0198] The collaborative dual-depth Q network algorithm uses the commonality of behaviors and goals between homogeneous mobile bodies to share neural network structures and experience sharing learning, simulating the information sharing of personnel on the development and spread of high temperature and toxic and harmful smoke in the real evacuation process through communication methods such as shouting. The shared learning method between mobile bodies enables the experience learned by a single mobile body at each step to be generalized to other mobile bodies, allowing each mobile body to learn by using the experience knowledge of other mobile bodies, improving the learning efficiency of the neural network and the speed of algorithm path planning. All mobile bodies can share a shared neural network for learning, and the number of neural networks and training parameters will not change with the increase or decrease of the number of mobile bodies, which enhances the stability of the evacuation model. Due to the sharing of the neural network structure and parameters of all mobile bodies, the difficulty of neural network training is reduced, and the learning efficiency of the neural network is further improved.
[0199] Furthermore, the state space of the moving object is set to S = {s1, s2, ..., s n}, represents the set of all states of the mobile in the environment, s t ∈S represents the state of the mobile body at time step t. Among them, n is the total number of state spaces, which means the number of all state spaces in which the mobile body can act except for obstacles in the environment.
[0200] Set the action space of the moving object A = {a1, a2, ..., a m}, represents the set of all the optional actions of the mobile in the current environment, a t ∈A represents the action of the mobile body at time step t, and m is the total number of action spaces.
[0201] Set the state transition probability function of the mobile body in the environment P:S×A×S→[0,1], P(s t+1 |s t ,a t )
[0202] Indicates that the moving body is in state s t Select action a t After that, the state changes to s t+1 The probability range is between 0 and 1.
[0203] Furthermore, a reward function for the moving body is set. Due to the large state space, in order to solve the sparse reward problem and speed up the model training, a negative reward is given to each action of the moving body, and a larger positive reward is given when the exit is reached. The reward is a type of the aforementioned safety benefit. The reward setting principles for different moving bodies are the same, and no joint reward for multiple moving bodies is set. The specific reward settings are as follows:
[0204]
[0205] Among them, A 1~4 ={a1,a2,a3,a4}, the action is to move one grid up, down, left, or right; A 5~8 ={a5,a6,a7,a8}, the action is to move one grid to the upper left, lower left, upper right, and lower right, and the moving distance is A 1~4 of times, so the reward is also set to A 1~4 of times.
[0206]
[0207]
[0208] Where n is the number of exports, d t and d t+1 are the distances between the moving body and the nearest exit at time step t and time step t+1 respectively.
[0209] The reward setting standard is the change in the distance between the moving body and the nearest exit. When the distance between the moving body and the nearest exit decreases, a smaller negative reward is given; when the distance between the moving body and the nearest exit does not decrease, a larger negative reward is given.
[0210] The reward for keeping the distance between the moving body and the nearest exit unchanged and increasing is set to be the same in order to encourage the moving body to explore other exit evacuation situations. In most cases, the unchanged distance between the moving body and the nearest exit indicates that the moving body is affected by fire products, obstacles or other people and cannot move in the selected direction and stays in place. Setting the same reward for keeping the distance unchanged and increasing encourages the moving body to go to other exits to try to get a better path planning result when the current exit is temporarily blocked by a large number of fire products or crowd congestion.
[0211] Furthermore, when the mobile body uses the collaborative dual-depth Q network algorithm to learn and train in the environment, the experience learned in several adjacent states is correlated and does not meet the independent and identically distributed requirements, which will cause the neural network training results to converge to a suboptimal strategy, or even cause the training results to oscillate and diverge and fail to converge. Therefore, the experience replay method is used to update the parameters of the neural network to improve the stability of the algorithm.
[0212] The experience replay mechanism is to replay the experience ancestor e after the mobile body makes an action selection. t =(s t ,a t ,r t ,s t+1 ) into the experience pool D = {e1, e2, ..., e N}, a part of the experience tuples in the experience pool is selected for learning by random batch sampling, and the neural network parameters θ are updated. In the collaborative dual-depth Q network algorithm, an experience tuple consists of four parameters, which are the current state s at time step t t , action a performed at time step t t , in state s t Execute action a t Rewards received t , and in state s t Execute action a t The next state to enter is s t+1 The number of experience tuples stored in the experience pool is limited. When the storage capacity of the experience pool reaches the upper limit, the newly added experience will replace the initially learned experience, so that the experience in the experience pool is updated in real time. In the process of neural network learning and training, the random batch sampling method reduces the correlation between experience tuples, enables the learning experience to be reused, and prevents the neural network from falling into the local optimum.
[0213] Step S350: Obtain the evacuation route and evacuation time of each person.
[0214] After the model training is completed, the evacuation path and evacuation time of each person can be obtained, as well as the distribution of fire hazard areas, the number of people evacuated, and the distribution of remaining people at any evacuation time. Figure 7 As shown, it is a schematic diagram of the end of personnel evacuation provided in an embodiment of the present application.
[0215] The embodiments of the present application take into account the real-time impact of fire burning on the evacuation of personnel, making the evacuation process of personnel safer and more realistic, and helping to plan a scientific and reasonable evacuation route for personnel.
[0216] The embodiment of the present application adopts a collaborative dual-depth Q network algorithm, which enables the mobile bodies to share neural networks, reduces the amount of neural network learning, and reduces the difficulty of training; it enables the mobile bodies to share learning experience, improves the learning efficiency of the neural network, and increases the speed of algorithm path planning.
[0217] The embodiment of the present application provides a fire evacuation simulation system based on deep reinforcement learning, including:
[0218] Initialize the setting module to initialize the personnel distribution, real-time fire danger zone, walls and safety exits in the scene model;
[0219] Neural network model building module, build collaborative deep reinforcement learning network model, set reward function for interaction with the environment, and build a large experience pool to store the experience of all mobile bodies;
[0220] The fire hazard area reading module obtains the real-time fire hazard area exceeding the temperature threshold or below the visibility threshold from the fire dynamics simulation software;
[0221] The evacuation simulation module uses a collaborative dual-depth Q-network algorithm to plan the evacuation path of personnel, obtain the optimal evacuation path planning result for the entire crowd, and obtain the evacuation path for each person separately.
[0222] The above are only preferred embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.
[0223] Although the above describes the specific implementation methods of the present application in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present application. Technical personnel in the relevant field should understand that on the basis of the technical solution of the present application, various modifications or deformations that can be made by technical personnel in this field without creative work are still within the scope of protection of the present application.
[0224] like Figure 8As shown, an embodiment of the present application provides an evacuation path determination device, the device comprising:
[0225] The generating module 110 is used to input the first model according to the building layout information and the accident information of the accident scene, and generate the accident map information at the current moment; wherein the accident map information at least indicates whether different areas of the accident scene are passable;
[0226] A first acquisition module 120 is used to acquire the distribution information of the accident scene targets and the exit information of the safety exit;
[0227] A first determination module 130, configured to determine relative position information between the target and the safety exit according to the distribution status information;
[0228] The obtaining module 140 is used to obtain a safe evacuation path for the target to leave the accident scene according to the accident map information, the relative position information and the second model.
[0229] In some embodiments, the generating module 110, the first acquiring module 120, the first determining module 130, and the obtaining module 140 may all be program modules, which can realize the functions of the above modules after being executed by a processor.
[0230] In other embodiments, the generation module 110, the first acquisition module 120, the first determination module 130, and the obtaining module 140 may all be soft-hard combination modules; the soft-hard combination modules include but are not limited to: various programmable arrays; the field programmable arrays include but are not limited to: field programmable arrays and / or complex programmable arrays.
[0231] In some other embodiments, the generating module 110, the first acquiring module 120, the first determining module 130, and the obtaining module 140 may all be pure hardware modules; the pure hardware modules include but are not limited to: application specific integrated circuits.
[0232] In some embodiments, the apparatus further comprises:
[0233] The second acquisition module is used to re-acquire the distribution status information of the target at the accident scene after the movement status of the target along the safe evacuation path meets the first condition.
[0234] In some embodiments, after the movement status of the target along the safe evacuation path satisfies the first condition, reacquiring the distribution status information of the target at the accident scene includes:
[0235] After the movement status of the target according to the safe evacuation path satisfies the first condition, determining whether the relative position relationship between the target and the safe exit satisfies a second condition;
[0236] If the relative position relationship between the target and the safety exit does not satisfy the second condition, the distribution status information of the target at the accident scene is re-acquired.
[0237] In some embodiments, the apparatus further comprises:
[0238] A stopping module is used to stop determining the safe evacuation path if the relative position between the target and the safe exit meets the second condition.
[0239] In some embodiments, obtaining a safe evacuation path for the target to leave the accident scene according to the accident map information, the relative position information and the second model includes:
[0240] Determining environmental information of the area where the target is located from the accident map information according to the distribution status information;
[0241] The environmental information and the relative position information are input into the second model to obtain a safe evacuation path for the target to leave the accident scene output by the second model.
[0242] In some embodiments, inputting the environmental information and the relative position information into the second model to obtain a safe evacuation path for the target to leave the accident site output by the second model includes:
[0243] Determining a plurality of alternative movement paths that meet safety conditions according to the accident map information and the relative position information;
[0244] When there are multiple alternative movement paths, the alternative movement path with the greatest safety benefit is selected as the safe evacuation path; wherein the safety benefit has one of the following specific characteristics:
[0245] The safety benefit is negatively correlated with the distance the target moves to the safety exit along the corresponding alternative moving path;
[0246] The safety benefit is negatively correlated with the accident rate predicted when the target moves to the safety exit according to the corresponding alternative moving path;
[0247] The safety benefit is positively correlated with the probability that the target successfully reaches the safety exit along the corresponding alternative moving path.
[0248] In some embodiments, the target distribution information indicates:
[0249] The distribution positions of the targets and / or the number of targets located within the accident scene;
[0250] The environmental information at least indicates the area type of the adjacent area where the target is located;
[0251] The area types include: a passable area and a non-passable area outside the passable area.
[0252] In some embodiments, the non-traffic area includes at least one of the following:
[0253] Obstacle area, an area where obstacles that hinder passage are piled up;
[0254] Accident area, the area where the accident occurred;
[0255] A congested area is an area that accommodates more than a preset number of targets.
[0256] In some embodiments, the apparatus further comprises:
[0257] A third acquisition module is used for the accident parameters of the accident scene;
[0258] The second determination module is used to determine whether the adjacent area of the target location is the accident area.
[0259] In some embodiments, the accident parameter includes at least one of the following:
[0260] Fire parameters;
[0261] Collapse parameters;
[0262] Flood parameters.
[0263] In some embodiments, the accident scene is a fire scene;
[0264] The obtaining of accident parameters at the accident scene includes:
[0265] The type of fire source at the fire scene, the spatial parameters of the space where the fire source is located, and the heat release rate of the fire source are obtained, wherein the spatial parameters of the space where the fire source is located indicate the ignition point position of the fire source and / or the distribution status of combustibles in the space where the fire source is located.
[0266] In some embodiments, determining whether an adjacent area of the target location is the accident area according to the accident parameter includes:
[0267] Input a third model according to the type of the fire source at the fire scene, the spatial parameters of the space where the fire source is located, and the heat release rate of the fire source to obtain an output result;
[0268] According to the output result, obtaining the temperature and / or visibility of each area of the fire scene;
[0269] Determining a dangerous area of the fire scene according to the temperature and / or visibility;
[0270] According to the dangerous area, it is determined whether an adjacent area where the target is located is the accident area.
[0271] In some embodiments, determining the dangerous area of the fire scene according to the temperature and / or visibility includes:
[0272] When the temperature exceeds a first threshold, and / or when the visibility is lower than a second threshold, the adjacent area where the target is located is determined to be a dangerous area of the fire scene.
[0273] like Fig. 9 As shown, an embodiment of the present application provides an electronic device, the electronic device comprising:
[0274] a memory for storing computer readable instructions;
[0275] A processor is connected to the memory and is used to implement the method provided by any of the above embodiments by executing computer-readable instructions, for example, the executable Figure 1 , Figure 2 and / or Figure 3 The method shown in .
[0276] The memory may be various types of memory, such as random access memory, read-only memory, flash memory, etc. The memory may be used for information storage, for example, storing computer executable instructions, etc. The computer executable instructions may be various program instructions, for example, target program instructions and / or source program instructions, etc.
[0277] The processor may be any type of processor, for example, a central processing unit, a microprocessor, a digital signal processor, a programmable array, a digital signal processor, an application specific integrated circuit or an image processor, etc. The processor may be connected to the memory via a bus, which may be an integrated circuit bus, etc.
[0278] like Fig. 9 As shown, the electronic device may also include a network interface, which can be used to interact with a peer device through a network.
[0279] The present application also provides a computer storage medium, which stores computer executable instructions. After the computer executable instructions are executed, the method provided in any of the above embodiments can be implemented. For example, the executable instructions Figure 1 , Figure 2 and / or Figure 3 The method shown in .
[0280] The computer storage medium provided in this embodiment includes: a mobile storage device, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.
[0281] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.
[0282] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.
[0283] In addition, all functional units in the embodiments of the present application may be integrated into one processing module, or each unit may be separately configured as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.
[0284] A person of ordinary skill in the art can understand that: all or part of the steps of implementing the above method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiment; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), disks or optical disks, etc. Various media that can store program codes.
[0285] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.
Claims
1. A method for determining an evacuation path, characterized in that: include: Inputting the first model according to the building layout information and accident information of the accident scene to generate the accident map information at the current moment; wherein the accident map information at least indicates whether different areas of the accident scene are accessible; the first model is an evacuation scene simulation model, which is used to generate the overall accident map information according to the input information; Obtaining distribution information of targets at the accident scene and exit information of emergency exits; Determining relative position information between the target and the safety exit according to the distribution status information; According to the accident map information, the relative position information and the second model, a safe evacuation path for the target to leave the accident scene is obtained; wherein the second model is a neural network model using a collaborative dual deep Q network algorithm as deep reinforcement learning; Wherein, obtaining the safe evacuation path for the target to leave the accident scene according to the accident map information, the relative position information and the second model includes: determining the environmental information of the area where the target is located from the accident map information according to the distribution status information; inputting the environmental information and the relative position information into the second model to obtain the safe evacuation path for the target to leave the accident scene output by the second model; The step of inputting the environmental information and the relative position information into the second model to obtain the safe evacuation path for the target to leave the accident scene output by the second model includes: determining a plurality of alternative movement paths that meet safety conditions according to the accident map information and the relative position information; when there are a plurality of the alternative movement paths, selecting the alternative movement path with the greatest safety benefit as the safe evacuation path; The safety benefit includes one of the following characteristics: the safety benefit is negatively correlated with the distance that the target moves to the safety exit along the corresponding alternative moving path; the safety benefit is negatively correlated with the accident rate predicted when the target moves to the safety exit along the corresponding alternative moving path; the safety benefit is positively correlated with the probability that the target successfully reaches the safety exit along the corresponding alternative moving path; The method further includes: after the movement status of the target according to the safe evacuation path satisfies a first condition, reacquiring distribution status information of the target at the accident site; wherein the first condition is the moving time or the moving distance of the target; the movement status satisfies the first condition including at least one of the following: the first condition is satisfied when the moving steps of the target reach a step number threshold; the first condition is satisfied when the moving time of the target reaches a time threshold; the first condition is satisfied when the distance traveled by the target reaches a distance length threshold; the first condition is satisfied when the displacement of the target reaches a displacement threshold; Wherein, after the movement status of the target according to the safe evacuation path satisfies the first condition, the distribution status information of the target at the accident scene is reacquired, including: after the movement status of the target according to the safe evacuation path satisfies the first condition, determining whether the relative position relationship between the target and the safe exit satisfies the second condition; if the relative position relationship between the target and the safe exit does not satisfy the second condition, reacquiring the distribution status information of the target at the accident scene; if the relative position between the target and the safe exit satisfies the second condition, stopping the determination of the safe evacuation path.
2. The method according to claim 1, characterized in that The target distribution information indicates: The distribution positions of the targets and / or the number of targets located within the accident scene; The environmental information at least indicates the area type of the adjacent area where the target is located; The area types include: a passable area and a non-passable area other than the passable area.
3. The method according to claim 2, characterized in that The non-traffic area includes at least one of the following: Obstacle area, an area where obstacles that hinder passage are piled up; Accident area, the area where the accident occurred; A congested area is an area that accommodates more than a preset number of targets.
4. The method according to claim 3, characterized in that The method further comprises: Acquiring accident parameters of the accident scene; According to the accident parameters, it is determined whether an adjacent area of the target location is the accident area.
5. The method according to claim 4, characterized in that The accident parameters include at least one of the following: Fire parameters; Collapse parameters; Flood parameters.
6. The method according to claim 4, characterized in that The accident scene is a fire scene; The obtaining of accident parameters at the accident scene includes: The type of fire source at the fire scene, the spatial parameters of the space where the fire source is located, and the heat release rate of the fire source are obtained, wherein the spatial parameters of the space where the fire source is located indicate the ignition point position of the fire source and / or the distribution status of combustibles in the space where the fire source is located.
7. The method according to claim 6, characterized in that The step of determining, based on the accident parameter, whether an adjacent area to the location of the target is the accident area comprises: Input a third model according to the type of the fire source at the fire scene, the spatial parameters of the space where the fire source is located, and the heat release rate of the fire source to obtain an output result; According to the output result, obtaining the temperature and / or visibility of each area of the fire scene; Determining a dangerous area of the fire scene according to the temperature and / or visibility; According to the dangerous area, it is determined whether an adjacent area where the target is located is the accident area.
8. The method according to claim 7, characterized in that Determining the dangerous area of the fire scene according to the temperature and / or visibility includes: When the temperature exceeds a first threshold, and / or when the visibility is lower than a second threshold, the adjacent area where the target is located is determined to be a dangerous area of the fire scene.
9. An evacuation path determination device, characterized in that: include: A generating module, configured to input the first model according to the building layout information and the accident information of the accident scene, and generate the accident map information at the current moment; wherein the accident map information at least indicates whether different areas of the accident scene are passable; A first acquisition module is used to acquire the distribution status information of the accident scene targets and the exit information of the safety exit; A first determining module, configured to determine relative position information between the target and the safety exit according to the distribution status information; The obtaining module is used to obtain a safe evacuation path for the target to leave the accident scene according to the accident map information, the relative position information and the second model.
10. An electronic device, characterized in that: include: a memory storing computer-readable instructions; A processor, connected to the memory, is configured to implement the method according to any one of claims 1 to 8 by executing the computer-readable instructions.
11. A computer storage medium, characterized in that: The computer storage medium stores computer executable instructions; after the computer executable instructions are executed by the processor, the method described in any one of claims 1 to 8 can be implemented.
