Unmanned boat emergency response method and system based on limited intervention dynamic programming

Abstract

The present invention relates to the technical field of emergency decision-making for unmanned boats, and specifically discloses an unmanned boat emergency handling method and system with limited intervention dynamic programming, the method comprising step 1: real-time monitoring of the unmanned boat, and generating alarm information when an abnormal situation occurs; step 2: generating a decision result according to the alarm information, sending the alarm information and the decision result to a remote end, and having the remote end manually determine whether to execute the decision result, if the decision result is executed, proceed to step 3, if not, proceed to step 1; step 3: determining a new route for the unmanned boat according to the decision result. The present invention generates a corresponding decision result according to the cause of the abnormal situation occurring in the unmanned boat, and determines whether the decision result is executed under limited manual intervention, thereby improving the efficiency and accuracy of emergency decision-making, effectively ensuring the safety of the unmanned boat in emergency situations, and having high use value.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned boat emergency decision-making, and in particular to an unmanned boat emergency handling method and system with limited intervention dynamic programming. Background Art

[0002] Due to the unmanned nature of unmanned boats, they need to make autonomous decisions in emergency situations, but the current level of intelligence cannot meet both accuracy and decision-making efficiency. There are three main problems with the current emergency decision-making of unmanned boats: first, the emergency decision-making plan is single and cannot cope with a variety of emergency conditions; second, it cannot achieve complete emergency planning and lacks the automatic route planning link; third, the accuracy of intelligent emergency decision-making is low and the decision-making efficiency is low. Summary of the invention

[0003] The present invention aims to solve the problem that the means of emergency response for unmanned boats are limited and inefficient. To this end, the present invention provides an unmanned boat emergency response method, system and product with limited intervention dynamic programming, which generates corresponding decision results according to the causes of abnormal situations occurring in the unmanned boat, and decides whether to execute the decision results under limited manual intervention, thereby improving the efficiency and accuracy of emergency decision-making, effectively ensuring the safety of unmanned boats in emergency situations, and having high use value.

[0004] The present invention provides an unmanned boat emergency response method with limited intervention dynamic programming, and the technical scheme adopted is as follows:

[0005] Step 1: Monitor the unmanned boat in real time and generate alarm information when an abnormal situation occurs;

[0006] Step 2: Generate a decision result according to the alarm information, send the alarm information and the decision result to the remote end, and the remote end manually determines whether to execute the decision result. If the decision result is executed, proceed to step 3; if not, proceed to step 1;

[0007] Step 3: Determine a new route for the unmanned boat based on the decision result.

[0008] Furthermore, the monitoring contents of the unmanned boat include cabins, unmanned boat intelligent cabins, navigation systems, radar monitoring systems, communication systems, navigation threats and environmental threats.

[0009] Furthermore, the cabin is monitored based on the cabin video. The specific process is: the cabin video is detected using the frame difference method. If environmental changes are detected, the target detection model is used to perform target detection on the image of the environmental changes. If the target detection result is smoke in the cabin, engine fire or personnel intrusion, it is determined that an abnormal situation has occurred and an alarm message is generated.

[0010] Furthermore, the target detection model includes five layers of Self-ONNs and five layers of reverse Self-ONNs connected in sequence, the five layers of Self-ONNs are respectively the first Self-ONNs, the second Self-ONNs, the third Self-ONNs, the fourth Self-ONNs and the fifth Self-ONNs, the five layers of reverse Self-ONNs are respectively the first reverse Self-ONNs, the second reverse Self-ONNs, the third reverse Self-ONNs, the fourth reverse Self-ONNs and the fifth reverse Self-ONNs, the first Self-ONNs is jump-connected to the fourth reverse Self-ONNs, the second Self-ONNs is jump-connected to the third reverse Self-ONNs, the third Self-ONNs is jump-connected to the second reverse Self-ONNs, and the fourth Self-ONNs is jump-connected to the first reverse Self-ONNs.

[0011] Furthermore, the process of determining the new route of the unmanned boat according to the decision result is as follows:

[0012] When the decision result is to continue the mission, the new route is the current route;

[0013] When the decision result is a new task, the new route is a pre-determined route;

[0014] When the decision result is a new route planning, the new route is a route planned according to the decision result and environmental monitoring information.

[0015] Furthermore, the calculation process of environmental monitoring information is:

[0016] The environment around the unmanned boat is detected to obtain target information. After the target information is spliced ​​with longitude, latitude and credibility, it is input into the deep learning model. The deep learning model assigns values ​​to the obstruction attributes of the target to obtain environmental monitoring information.

[0017] Furthermore, the process of planning the route according to the decision result and the environmental monitoring information is as follows:

[0018] Step 31: Determine the destination and planning strategy according to the decision result; the planning strategy includes time distance priority and physical distance priority;

[0019] Step 32: Select a chart area based on the current position and destination of the unmanned boat;

[0020] Step 33: Adaptively perform grid division according to the chart area and environmental monitoring information to obtain a grid division result;

[0021] Step 34: Generate an obstruction attribute for each grid according to the grid division result, the chart area and the environmental monitoring information to obtain an obstruction attribute matrix;

[0022] Step 35: Obtain a new route based on the obstruction attribute matrix and planning strategy.

[0023] Further, according to the alarm information, a decision result is generated using a rule base, and the rule base is shown in Table 1:

[0024] Table 1 Rule library table

[0025] .

[0026] The present invention also provides an unmanned boat emergency response system with limited intervention dynamic programming, which adopts the following technical solution: comprising: an abnormality detection module, an intelligent decision-making module and a route dynamic planning module connected in sequence,

[0027] The abnormality detection module is used to monitor the unmanned boat in real time and generate alarm information when an abnormal situation occurs;

[0028] The intelligent decision-making module is used to generate a decision result according to the alarm information, send the alarm information and the decision result to a remote end, and the remote end manually determines whether to execute the decision result. If the decision result is executed, the decision result is sent to the route dynamic planning module;

[0029] The route dynamic planning module is used to determine a new route for the unmanned boat according to the decision result.

[0030] It also includes an environmental monitoring module, which is connected to the route dynamic planning module, and the environmental monitoring module is used to generate environmental monitoring information.

[0031] The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements the above-mentioned limited intervention dynamic programming unmanned boat emergency response method.

[0032] The above one or more technical solutions in the embodiments of the present invention have at least one of the following technical effects:

[0033] 1. The present invention organically combines emergency decision-making with route planning, strictly closes the alarm information, and enhances the maneuverability and flexibility of the unmanned boat when performing tasks. The present invention introduces limited human intervention to determine whether the decision result is executed, ensures the accuracy of emergency decision-making, and combines the route planning algorithm to complete the final link of the emergency decision-making plan.

[0034] 2. The present invention builds a target detection model based on operational U-Nets and uses Self-ONNs to quickly detect human intrusion and fire. Compared with the traditional YOLO, the target detection model combines detection and calibration anchor frames together, which improves the generation speed and is more suitable for unmanned scenes.

[0035] 3. The present invention has formulated corresponding appropriate decision results for various alarm information. In addition, the route planning also takes various factors into consideration, takes into account environmental monitoring information, and provides two optional distance calculation methods, which can adaptively generate multiple strategy route planning results for the unmanned boat, and has high application value.

[0036] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0038] Figure 1 It is a flow chart of the method provided by the present invention.

[0039] Figure 2 It is a structural schematic diagram of the target detection module provided by the present invention.

[0040] Figure 3 It is a schematic diagram of route planning distance calculation provided by the present invention.

[0041] Figure 4 It is a structural block diagram of the system provided by the present invention.

[0042] Reference numerals:

[0043] 1. Anomaly detection module; 2. Intelligent decision-making module; 3. Dynamic route planning module; 4. Environmental monitoring module. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical scheme and advantages of the present invention clearer, the technical scheme of the present invention will be clearly and completely described below in conjunction with the drawings in the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present invention. The following embodiments are used to illustrate the present invention, but cannot be used to limit the scope of the present invention.

[0045] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the embodiment of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0046] Combine the following Figures 1 to 4 The present invention is further described in detail, and a limited intervention dynamic programming unmanned boat emergency response method, system and product of the present invention are described:

[0047] In this embodiment, Figure 1 As shown, a limited intervention dynamic programming unmanned boat emergency response method is provided, comprising the following steps:

[0048] Step 1: Monitor the unmanned boat in real time and generate alarm information when an abnormal situation occurs.

[0049] The monitoring contents of unmanned boats include cabins, intelligent engine rooms, navigation systems, radar monitoring systems, communication systems, navigation threats and environmental threats. Among them, navigation threats refer to encountering targets that hinder the safe navigation of unmanned boats, including unavoidable targets, water ingress, etc., and environmental threats refer to encountering external attacks and high sea conditions warnings (typhoon weather and sea conditions that are not suitable for continued navigation, etc.). When any monitoring content of an unmanned boat is abnormal, an alarm message is generated, and the alarm message includes the specific monitoring content where the abnormal situation occurred.

[0050] According to the monitoring content of the unmanned boat where the abnormal situation occurs, the alarm information can be divided into three categories, namely fire / intrusion alarm, health monitoring alarm and threat level alarm. In this embodiment, the fire / intrusion alarm includes cabin smoke, engine fire and personnel intrusion, which correspond to the cabin in the monitoring content. Health monitoring alarms include low oil level, navigation failure, radar failure and communication failure, which correspond to the unmanned boat intelligent cabin, navigation system, radar monitoring system and communication system in the monitoring content respectively. Threat level alarms include unavoidable collision targets, water ingress, high sea condition warnings and external attacks, which correspond to navigation threats and environmental threats in the unmanned boat.

[0051] Taking into account the situation of security and fire fighting services, and combining the unmanned characteristics of unmanned boats, the present invention adopts the frame difference method to monitor the information of changes in the cabin environment. If changes in the environment are detected, target detection is performed to determine whether there is human intrusion or fire, so as to reduce the system load and improve the recognition accuracy.

[0052] The specific process of cabin-based video monitoring is as follows: the frame difference method is used to detect the cabin video. If environmental changes are detected, the target detection model is used to perform target detection on the image of the environmental changes. If the target detection result is smoke in the cabin, engine fire or personnel intrusion, it is determined that an abnormal situation has occurred and an alarm message is generated.

[0053] like Figure 2As shown, the target detection model is constructed based on operational U-Nets, including five layers of Self-ONNs (Self-organized operational neural networks) and five layers of reverse Self-ONNs connected in sequence, the five layers of Self-ONNs are respectively the first Self-ONNs, the second Self-ONNs, the third Self-ONNs, the fourth Self-ONNs and the fifth Self-ONNs, and the five layers of reverse Self-ONNs are respectively the first reverse Self-ONNs, the second reverse Self-ONNs, the third reverse Self-ONNs, the fourth reverse Self-ONNs and the fifth reverse Self-ONNs. The first Self-ONNs, the second Self-ONNs, the third Self-ONNs, the fourth Self-ONNs, the fifth Self-ONNs, the first reverse Self-ONNs, the second reverse Self-ONNs, the third reverse Self-ONNs, the fourth reverse Self-ONNs and the fifth reverse Self-ONNs are connected in sequence, the first Self-ONNs are jump-connected to the fourth reverse Self-ONNs, the second Self-ONNs are jump-connected to the third reverse Self-ONNs, the third Self-ONNs are jump-connected to the second reverse Self-ONNs, and the fourth Self-ONNs are jump-connected to the first reverse Self-ONNs.

[0054] The parameters of each layer of the target detection model are as follows: the first Self-ONNs, N: 8, KS: 5, S: 2, Q: 3; the second Self-ONNs, N: 16, KS: 5, S: 2, Q: 3; the third Self-ONNs, N: 64, KS: 5, S: 2, Q: 3; the fourth Self-ONNs, N: 128, KS: 5, S: 2, Q: 3; the fifth Self-ONNs, N: 512, KS: 5, S: 2, Q: 3; the first reverse Self-ONNs, N :512, KS:5, S:2, Q:3; the second reverse Self-ONNs, N:128, KS:5, S:2, Q:3; the third reverse Self-ONNs, N:64, KS:5, S:2, Q:3; the fourth reverse Self-ONNs, N:32, KS:5, S:2, Q:3; the fifth reverse Self-ONNs, N:8, KS:5, S:2, Q:3; where N represents the number of channels, KS represents the kernel size, S represents the sampling factor, and Q represents the order of nonlinear transformation.

[0055] Five layers of Self-ONNs constitute the encoding module, which is used to extract useful features in the image; five layers of reverse Self-ONNs constitute the decoding module, which is used to generate semantic segmentation boxes using the learned features; the jump connections between Self-ONNs are used to splice features, thereby indirectly helping the formation of gradient flow.

[0056] Compared with neural networks such as CNN and ONN, self-onns greatly increases nonlinear capabilities, avoids the problem of parameter surge caused by network complexity, and can greatly improve the generation quality of semantic segmentation boxes. The skip connection method adopted can improve the mapping capability of the entire network framework.

[0057] When training the target detection model, after detecting environmental changes using the frame difference method, the target detection model is called to perform target detection, determine whether there is a human intrusion or a fire, and generate alarm information. Then, it is manually determined whether a false alarm is generated. If a false alarm is generated, the correct data is marked and stored in the database, and batch retraining is automatically performed to optimize the target detection model.

[0058] Step 2: Generate a decision result based on the alarm information, send the alarm information and decision result to the remote end, and let the remote end manually determine whether to execute the decision result. If the decision result is executed, go to step 3; if not, go to step 1.

[0059] According to the alarm information, a decision result is generated using a rule base. In this embodiment, the rule base is shown in Table 1.

[0060] Table 1 Rule library table

[0061]

[0062] The decision results are divided into three types, including continuing the mission, new mission and new route planning.

[0063] Continuing the mission means that the unmanned boat continues to perform the current mission, that is, continues to navigate normally along the current route.

[0064] The new mission is for the unmanned boat to sail according to the pre-determined route and no longer continue the current mission. Specifically, sail at full rudder and indicate the hazard warning light: the unmanned boat is actually sailing in circles, waiting for rescue, turning on the hazard warning light to alert nearby ships. In the event of water ingress during navigation, the unmanned boat can also close the internal partition door to isolate the flooded area. Return according to the predetermined plan: return according to the predetermined plan, and return normally. Escape route: sail according to the predetermined escape route.

[0065] The new route planning sets the destination and planning strategy, and it is necessary to dynamically plan the new route of the unmanned boat in combination with the environmental monitoring information. Table 1 gives the specific contents of the four new route planning. The destinations include the nearest coast (nearest coast to dock), the departure point of the unmanned boat (return) and the original destination of the unmanned boat (continue to perform the mission); the planning strategies include: time distance priority and physical distance priority.

[0066] Whether the decision result is executed requires human intervention, that is, human confirmation of the decision result to decide whether to execute the decision result.

[0067] For the rule base, entries can be added, deleted, or modified according to actual situations. The manual confirmation of the decision results can also be used as a reference to determine whether the rule base needs to be updated.

[0068] Step 3: Determine a new route for the unmanned boat based on the decision result, and plan the route of the unmanned boat with limited intervention.

[0069] The process of determining the new route of the unmanned boat according to the decision result is as follows:

[0070] When the decision result is to continue the mission, the new route is the current route;

[0071] When the decision result is a new task, the new route is a pre-determined route;

[0072] When the decision result is a new route planning, the new route is a route planned according to the decision result and environmental monitoring information.

[0073] The calculation process of environmental monitoring information is as follows:

[0074] Radar and other devices are used to detect the environment around the unmanned boat to obtain target information. After the target information is spliced ​​with longitude, latitude and credibility, it is input into the deep learning model. The deep learning model assigns values ​​to the obstruction attributes of the target to obtain environmental monitoring information. The obstruction attributes can be manually set with initial values ​​and optimized during the iteration process of the deep learning model. Target information can include rocks, floating objects, etc.

[0075] The process of planning a route based on the decision results and environmental monitoring information is as follows:

[0076] Step 31: Determine the destination and planning strategy according to the decision result.

[0077] Step 32: Select a chart area based on the current location and destination of the unmanned boat.

[0078] Step 33: Adaptively perform grid division according to the chart area and environmental monitoring information to obtain a grid division result; the division process considers the minimum grid distance threshold, which can be set manually.

[0079] Step 34: Generate an obstruction attribute for each grid according to the grid division result, the chart area and the environmental monitoring information to obtain an obstruction attribute matrix.

[0080] Specifically, the default obstruction attribute of the grid is 0.5. The obstruction objects on the chart are mapped to the corresponding grids, and their obstruction attribute is 1. The channel is an optional configuration. If the channel is adapted, the obstruction attribute of the grid corresponding to the channel is 0.1. The obstruction attribute of the harmful targets in the environmental monitoring information is increased according to the damage range, with the highest being 0.9 and the lowest being 0.5.

[0081] Step 35: Obtain a new route based on the obstruction attribute matrix and planning strategy.

[0082] Select the corresponding distance calculation method according to the planning strategy. The distance calculation method is divided into two types: physical distance calculation method and time distance calculation method. Time distance priority corresponds to the time distance calculation method; physical distance priority corresponds to the physical distance calculation method.

[0083] When using the distance calculation method to calculate a new route, grids with an obstruction attribute of 1 are not taken into account, and alternative routes are determined. The physical distance (or time distance) of the alternative routes is calculated, and the physical distance calculation result (or time distance calculation result) is multiplied by the average obstruction attribute of the grid corresponding to the alternative route. The route with the smallest result value is selected as the new route.

[0084] Assume that the new route is calculated using the time distance calculation method. Figure 3 As shown in the figure, from point A (current position of the unmanned boat) to point B (destination), the obstruction attribute of the channel is 0.1, the obstruction attribute of the obstacle is 1, the obstruction attribute of the environmental factor is 0.9, and the obstruction attribute of the remaining grids is 0.5. The red, green, and yellow alternative routes are all reachable (the red and yellow share three grids). The yellow alternative route is the most suitable route, followed by the red alternative route, and the green alternative route is the third. The time distance calculation results of the three are 8, 9, and 10 respectively. The time distance calculation result is multiplied by the average value of the obstruction attribute of the grid corresponding to the alternative route. The result of the yellow alternative route has the smallest value and is the new route.

[0085] Although the time-distance calculation result of the red alternative route is smaller than that of the green alternative route, its average value of obstruction attribute is much higher than that of the green alternative route, and the final calculation result is also higher than that of the green alternative route.

[0086] In this embodiment, Figure 4As shown, an unmanned boat emergency response system with limited intervention dynamic planning is also provided, and the technical solution adopted is as follows: it includes: an anomaly detection module 1, an intelligent decision-making module 2, a route dynamic planning module 3 and an environmental monitoring module 4, the anomaly detection module 1 is connected to the intelligent decision-making module 2, and the route dynamic planning module 3 is connected to the intelligent decision-making module 2 and the environmental monitoring module 4.

[0087] The abnormality detection module 1 is used to monitor the unmanned boat in real time and generate alarm information when an abnormal situation occurs.

[0088] The monitoring contents of the unmanned boat include cabins, intelligent engine rooms of the unmanned boat, navigation systems, radar monitoring systems, communication systems, navigation threats and environmental threats.

[0089] The abnormality detection module 1 is divided into three submodules, namely, a fire / intrusion detection submodule, a health status monitoring submodule, and a threat level monitoring submodule. Each submodule is responsible for monitoring a part of the monitoring content of the unmanned boat. Specifically, the fire / intrusion detection submodule is used to monitor the cabin. When an abnormal situation occurs, a fire / intrusion alarm is generated. The fire / intrusion alarm includes cabin smoke, engine fire, and personnel intrusion; the health status monitoring submodule is used to monitor the unmanned boat's intelligent cabin, navigation system, radar monitoring system, and communication system. When an abnormal situation occurs, a health monitoring alarm is generated. The health monitoring alarm includes low oil level, navigation failure, radar failure, and communication failure; the threat level monitoring submodule is used to monitor navigation threats and environmental threats. When an abnormal situation occurs, a threat level alarm is generated. The threat level alarm includes unavoidable collision targets, water ingress, high sea level warnings, and external strikes.

[0090] Taking into account the situation of security and firefighting services, and combined with the unmanned boat's unmanned characteristics, the same detection method is used for fire and intrusion. Based on the video monitoring of the cabin, the specific process is: the video of the cabin is detected by the frame difference method. If an environmental change is detected, the target detection model is used to detect the target of the image of the environmental change. If the target detection result is smoke in the cabin, the engine is on fire, or a person is intruding, it is determined that an abnormal situation has occurred and an alarm message is generated. The target detection model is constructed based on an operational U-Nets, including five layers of Self-ONNs and five layers of reverse Self-ONNs connected in sequence. The five layers of Self-ONNs are the first Self-ONNs, the second Self-ONNs, the third Self-ONNs, the fourth Self-ONNs and the fifth Self-ONNs, and the five layers of reverse Self-ONNs are the first reverse Self-ONNs, the second reverse Self-ONNs, the third reverse Self-ONNs, the fourth reverse Self-ONNs and the fifth reverse Self-ONNs. The first Self-ONNs are jump-connected to the fourth reverse Self-ONNs, the second Self-ONNs are jump-connected to the third reverse Self-ONNs, the third Self-ONNs are jump-connected to the second reverse Self-ONNs, and the fourth Self-ONNs are jump-connected to the first reverse Self-ONNs.

[0091] When any of the fire / intrusion detection submodule, the health status monitoring submodule, and the threat level monitoring submodule detects an abnormal situation, an alarm message will be generated and then sent to the intelligent decision-making module 2.

[0092] The intelligent decision-making module 2 is used to generate a decision result based on the alarm information, and send the alarm information and decision result to a remote end, which manually determines whether to execute the decision result. If the decision result is to be executed, the decision result is sent to the route dynamic planning module 3.

[0093] The intelligent decision-making module 2 generates a decision result using a rule base according to the alarm information. The decision results are divided into three types, including continued tasks, new tasks and new route planning. After being sent to the remote end, the remote personnel decide whether to execute the decision result based on the alarm information and the decision result. The present invention introduces limited manual intervention to determine whether the decision result is executed, thereby ensuring the accuracy of emergency decisions. The remote end manually determines to execute the decision result, and the intelligent decision-making module 2 sends the decision result to the route dynamic planning module 3.

[0094] The route dynamic planning module 3 is used to determine the new route of the unmanned boat according to the decision result. The route dynamic planning module 3 analyzes the decision result and determines the new route of the unmanned boat according to the decision result. When the decision result is to continue the task, the new route is the current route; when the decision result is a new task, the new route is a pre-determined route; when the decision result is a new route planning, the environmental monitoring module 4 is called to obtain the environmental monitoring information generated by it, and a new route is planned according to the decision result and the environmental monitoring information.

[0095] The environment monitoring module 4 is used to generate environment monitoring information and then send the environment monitoring information to the route dynamic planning module 3.

[0096] In this embodiment, a computer program product is also provided, including a computer program, which, when executed by a processor, implements the above-mentioned limited intervention dynamic programming unmanned boat emergency response method.

[0097] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An unmanned boat emergency response method with limited intervention dynamic programming, characterized in that: include: Step 1: Monitor the unmanned boat in real time and generate alarm information when an abnormal situation occurs; The monitoring contents of the unmanned boat include cabins, intelligent engine rooms of the unmanned boat, navigation systems, radar monitoring systems, communication systems, navigation threats and environmental threats; The specific process of cabin-based video monitoring is as follows: the frame difference method is used to detect the cabin video. If an environmental change is detected, the target detection model is used to detect the target of the image of the environmental change. If the target detection result is smoke in the cabin, engine fire or personnel intrusion, it is determined that an abnormal situation has occurred and an alarm message is generated; The target detection model includes five layers of Self-ONNs and five layers of reverse Self-ONNs connected in sequence, the five layers of Self-ONNs are respectively the first Self-ONNs, the second Self-ONNs, the third Self-ONNs, the fourth Self-ONNs and the fifth Self-ONNs, the five layers of reverse Self-ONNs are respectively the first reverse Self-ONNs, the second reverse Self-ONNs, the third reverse Self-ONNs, the fourth reverse Self-ONNs and the fifth reverse Self-ONNs, the first Self-ONNs is jump-connected to the fourth reverse Self-ONNs, the second Self-ONNs is jump-connected to the third reverse Self-ONNs, the third Self-ONNs is jump-connected to the second reverse Self-ONNs, and the fourth Self-ONNs is jump-connected to the first reverse Self-ONNs; Step 2: Generate a decision result according to the alarm information, send the alarm information and the decision result to the remote end, and the remote end manually determines whether to execute the decision result. If the decision result is executed, proceed to step 3; if not, proceed to step 1; According to the alarm information, a decision result is generated using a rule base, and the rule base is shown in Table 1: Table 1 Rule library table Step 3: Determine a new route for the unmanned boat according to the decision result; The process of determining the new route of the unmanned boat according to the decision result is as follows: When the decision result is to continue the mission, the new route is the current route; When the decision result is a new task, the new route is a pre-determined route; When the decision result is a new route planning, the new route is a route planned according to the decision result and environmental monitoring information; The calculation process of environmental monitoring information is as follows: Detect the environment around the unmanned boat and obtain target information. After the target information is combined with the latitude, longitude and credibility, it is input into the deep learning model. The deep learning model assigns the obstruction attribute of the target to obtain environmental monitoring information. The process of planning a route based on the decision results and environmental monitoring information is as follows: Step 31: Determine the destination and planning strategy according to the decision result; the planning strategy includes time distance priority and physical distance priority; Step 32: Select a chart area based on the current position and destination of the unmanned boat; Step 33: Adaptively perform grid division according to the chart area and environmental monitoring information to obtain a grid division result; Step 34: Generate an obstruction attribute for each grid according to the grid division result, the chart area and the environmental monitoring information to obtain an obstruction attribute matrix; Step 35: Obtain a new route based on the obstruction attribute matrix and planning strategy.

2. An unmanned boat emergency response system with limited intervention dynamic programming, characterized in that: The method for emergency handling of an unmanned boat with limited intervention dynamic programming as claimed in claim 1 comprises: an abnormality detection module, an intelligent decision-making module and a route dynamic planning module connected in sequence, The abnormality detection module is used to monitor the unmanned boat in real time and generate alarm information when an abnormal situation occurs; The intelligent decision-making module is used to generate a decision result according to the alarm information, send the alarm information and the decision result to a remote end, and the remote end manually determines whether to execute the decision result. If the decision result is executed, the decision result is sent to the route dynamic planning module; The route dynamic planning module is used to determine a new route for the unmanned boat according to the decision result.

3. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the unmanned boat emergency response method of limited intervention dynamic programming described in claim 1 is implemented.

Images