Multi-mode autonomous perception and integrated management system of intelligent ship

Abstract

The invention provides a multi-mode autonomous perception and integrated management system of an intelligent ship, and belongs to the technical field of intelligent ships. The system comprises a path planning module, a fault detection module, a passenger service module and a safety evacuation module. The path planning module performs path planning according to the ship environment data acquired in real time; the fault detection module performs real-time fault detection according to the ship environment data; the passenger service module comprises communication equipment for collecting customer service requests and a robot for providing services; constructing an optimization model by taking the minimum total service waiting time as a target to generate a behavior instruction of each robot; and the safe evacuation module is used for dynamically generating an optimal evacuation path by fusing a fire behavior sensor, a water immersion sensor and passenger position data. According to the system, a passenger service module uses communication equipment and service robots to construct an optimization model by taking the minimum total service waiting time as a target to generate a behavior instruction of each robot; and orderly and efficient passenger service is realized.

Description

Technical Field

[0001] This invention relates to the field of intelligent ship technology, and in particular to a multimodal autonomous perception and integrated management system for intelligent ships. Background Technology

[0002] As the core carrier of modern shipping technology, intelligent ships require integrated management systems that simultaneously address multi-dimensional challenges related to the navigation environment, safety warnings, and passenger services. In complex maritime operations, ships need to process real-time meteorological and hydrological data as well as heterogeneous information streams of passenger behavior characteristics. These two types of data correspond to the two core functions defined by the International Maritime Organization (IMO): navigation safety and personnel management. Traditional ship management systems typically employ an independent subsystem architecture, resulting in inefficient data interaction and a lack of a unified decision support platform. The environmental perception data and passenger service demand data generated during ship navigation exhibit significant spatiotemporal correlations, necessitating a system with multi-source information fusion and collaborative processing capabilities.

[0003] Existing technical solutions suffer from the following limitations: First, the navigation environment monitoring system and passenger service system are designed separately, preventing data exchange during emergency response. Second, the path planning algorithm used in the safety early warning module has a trade-off between computational speed and accuracy, making it difficult to meet real-time requirements. Finally, the passenger service module lacks a dynamic optimization mechanism, resulting in a significant drop in service efficiency during emergencies. Particularly in high-density passenger areas, existing systems cannot simultaneously meet the dual requirements of optimal evacuation route calculation and rational allocation of service resources. These technical deficiencies severely restrict the overall performance of the intelligent ship's multimodal autonomous perception and integrated management system under complex operating conditions.

[0004] Therefore, a multimodal autonomous perception and integrated management system for intelligent ships is needed. Summary of the Invention

[0005] In view of this, the present invention provides a multimodal autonomous perception and integrated management system for intelligent ships, which performs path planning based on real-time collected ship environmental data; utilizes communication equipment for collecting customer service requests and robots for providing services; generates behavioral instructions for each robot by constructing an optimization model with the goal of minimizing total service waiting time; and dynamically generates the optimal evacuation route by integrating fire sensor, flood sensor and passenger location data; thereby achieving integrated management of the ship.

[0006] Therefore, the present invention provides the following technical solution: A multimodal autonomous perception and integrated management system for intelligent ships includes: The system includes a route planning module, a fault detection module, a passenger service module, and a safety evacuation module. The path planning module performs path planning based on real-time collected ship environmental data; The fault detection module performs real-time fault detection based on ship environmental data. The passenger service module includes: Communication devices that collect customer service requests and robots that provide services; an optimization model is built with the goal of minimizing the total service wait time to generate behavioral instructions for each robot; The safety evacuation module integrates fire sensors, water immersion sensors, and passenger location data to dynamically generate the optimal evacuation route.

[0007] Furthermore, the route planning based on real-time collected ship environmental data includes: The system collects real-time data on the ship's surrounding environment using sensors, including the location of static obstacles, the trajectory and speed of dynamic obstacles, typhoon paths, and meteorological data. Multiple initial feasible paths connecting the starting point and the target point are generated based on the A* algorithm, where the path nodes satisfy the minimum turning radius and speed constraints of the ship. The initial path is subjected to spatiotemporal conflict detection based on the predicted trajectory of dynamic obstacles, and the path is corrected by combining the collision avoidance priority in the COLREGs rules. The correction includes: When encountering a vessel approaching from the starboard side, turn to the starboard to avoid it; when encountering a vessel crossing from the port side, maintain course and reduce speed. When the direction of typhoon movement blocks the shipping channel, the detour route to the left of the typhoon's path is given priority. If this is not feasible, an anchorage waiting route is generated. Each candidate path undergoes a multi-objective evaluation, including: Calculate the efficiency score and the safety score of minimum obstacle distance under the fuel consumption model, and generate the comprehensive optimal path through a weighted function; Output the optimal path and dynamically replan during the voyage by periodically updating the data.

[0008] Furthermore, the total service wait time:

[0009] in, This indicates the total number of service requests. Indicates the first The arrival time of each request, This indicates that the robot has begun processing the first... The time of each request , No. The waiting time for each request. Total service wait time.

[0010] Furthermore, the optimization model also includes constraints: Robot single-task constraint: A robot can only perform one task at any given time; Task completion constraint: Each service request must be assigned to one and only one robot; Physical sequence constraint: The robot must execute tasks in the order they appear in its task queue, and the start time of a later task must not be earlier than the end time of the previous task. Movement time constraint: The robot is for the request The service start time shall not be earlier than the sum of the request arrival time and the time required for the robot to move to the requested location.

[0011] Furthermore, the optimization model is solved to generate behavioral instructions for each robot: All service robot task queues Initialize to empty; Record the next idle time of all robots. ; According to the requested predicted arrival time Sequential processing, for the current request Iterate through each robot : Calculate the estimated start time: ;in, For robots Move to request Time required at the location; calculate estimated waiting time ; Request Assigned to enable The smallest robot ,in ; Request Join the robot Task queue At the end, and update the robot. Next free time: , It is a service request. The time required for the event itself.

[0012] Furthermore, the behavioral instructions are: Task allocation matrix A binary matrix that defines which robot serves which request; Task queue collection : ,in It is an ordered list containing the assignments to the robots. All tasks are arranged in the order of execution; Each robot operates according to its own unique and ordered system. The task queue executes tasks sequentially.

[0013] Furthermore, the fusion of fire sensor, water immersion sensor, and passenger location data to dynamically generate the optimal evacuation route includes: The evaluation function is:

[0014] in, From the starting point to the node The actual cost; For the node The estimated cost to reach the destination; For nodes The total estimated cost; This is a weighting coefficient, dynamically adjusted according to the urgency of the emergency; when the emergency is more urgent, it is increased. The value of makes the algorithm more inclined to choose the shorter path; when the emergency is relatively mild, the value of is reduced. The value is determined by comprehensively considering the safety and distance of the path.

[0015] Furthermore, the real-time fault detection based on ship environmental data includes: The vibration sensor collects the vibration signal of the engine in real time. When an abnormal peak value appears within the preset frequency range, it is judged to be a bearing wear fault. Temperature sensors monitor temperature changes in the engine room. When the temperature sensor reading exceeds a preset temperature threshold, it is determined that the engine room is overheating. The pressure sensor monitors the pressure in the oil pipeline, and when the pressure sensor reading is lower than the preset pressure threshold, it is determined to be a fuel leak.

[0016] Advantages and positive effects of the present invention: The passenger service module of this system utilizes communication equipment and service robots to generate behavioral instructions for each robot by constructing an optimization model with the goal of minimizing the total service waiting time; thus achieving orderly and efficient passenger service.

[0017] The path planning module integrates dynamic weather obstacle trajectories and speeds, typhoon paths and meteorological data, as well as navigation rules to plan paths, providing more reliable paths. The safety evacuation module dynamically adjusts the weights of the heuristic function in the evaluation function based on the urgency of emergencies such as fire and flooding, achieving a balance between safety and efficiency. In the event of fire or flooding, dangerous routes are dynamically closed and alternative routes are activated to ensure the real-time optimality of evacuation routes. Safety factors are incorporated into the cost function to prevent passengers from being guided through dangerous areas. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a framework diagram of the multimodal autonomous perception and integrated management system for intelligent ships in an embodiment of the present invention; Figure 2 This is a timing diagram of the dynamic update of passenger evacuation routes in an embodiment of the present invention. Detailed Implementation

[0020] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0021] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0022] This invention provides a multimodal autonomous perception and integrated management system for intelligent ships, including: a path planning module, a fault detection module, a passenger service module, and a safety evacuation module; the path planning module performs path planning based on real-time collected ship environmental data; the passenger service module includes communication devices for collecting customer service requests and robots for providing services; an optimization model is constructed with the goal of minimizing the total service waiting time to generate behavioral instructions for each robot; the safety evacuation module integrates fire sensors, water immersion sensors, and passenger location data to dynamically generate the optimal evacuation route.

[0023] Example 1 The route planning module performs route planning based on real-time collected ship environmental data: Navigation information and shipboard environment information are collected through multiple source sensors.

[0024] Navigation information is obtained by collecting real-time data from lidar, radar, automatic identification system, wind speed, and wave height sensors.

[0025] A digital map of the navigation environment is constructed by combining distance information collected by lidar and radar with ambient propagation information collected by the Automatic Identification System (AIS) and data from wind speed and wave height sensors. 1) The distance data measured by the lidar is The distance data measured by the radar is The vessel location information provided by the Automatic Identification System (AIS) is The wind speed measured by the meteorological sensor is The wave height is The merged navigation environment information:

[0026] in, This is a preset data fusion function.

[0027] 2) By processing lidar and radar data in real time and employing target detection algorithms, preferably the YOLO model in this embodiment, other dynamic obstacles such as ships and floating objects are identified and their trajectories are tracked: Obstacles at any time The position is ,speed The next moment The predicted location is According to the kinematic formula:

[0028]

[0029] Based on historical and real-time meteorological data collected by meteorological sensors, and using meteorological prediction models, this embodiment preferably uses a numerical weather prediction model to predict meteorological risks such as typhoon paths.

[0030] Historical meteorological data is Real-time weather data is If the meteorological forecasting model is M, then the predicted typhoon path is: .

[0031] Obstacle avoidance is performed based on navigation environment information, predicted obstacle trajectories, and predicted typhoon paths.

[0032] The A algorithm is used to generate multiple feasible paths connecting the starting point and the target point. At the same time, considering the constraints of ship motion, the paths are modified according to the predicted obstacle trajectories to avoid conflicts and ensure compliance with the COLREGs rules.

[0033] Prioritize avoiding the right side of the typhoon's movement direction. If the typhoon's path blocks the shipping lanes, plan an alternative route or anchor and wait.

[0034] Calculate the efficiency score and the safety score of minimum obstacle distance under the fuel consumption model, and generate the comprehensive optimal path through a weighted function.

[0035] Let the fuel efficiency target be The safe distance target is Then the multi-objective optimization problem can be expressed as:

[0036] Where 'a' represents the collision avoidance action. and These are the weighting coefficients for the two objectives, respectively.

[0037] Output: heading, speed commands, or emergency braking signals.

[0038] Example 2 The system collects vibration signals from the engine, temperature signals from the engine room, and pressure signals from the fuel lines to monitor the health status of the equipment. Vibration sensors collect engine vibration signals in real time and use methods such as spectrum analysis to detect fault characteristics such as bearing wear; temperature sensors monitor temperature changes in the engine room to determine if there are any abnormalities such as overheating; and pressure sensors monitor the pressure in the fuel lines to detect problems such as fuel leaks.

[0039] 1) The spectrum of the vibration signal is When in a specific frequency range When abnormal peak values ​​appear within the bearing, it can be determined that the bearing is worn. When the temperature measured by the temperature sensor is greater than the preset temperature threshold, it is determined that the engine room is overheating. The pressure sensor measures the pressure as P. When the measured pressure value is less than the preset pressure threshold, it is determined to be a fuel leak.

[0040] 2) Train a long short-term memory network model using the equipment's historical operating data, and use the trained model to predict the equipment's future operating status based on the collected operating data, thereby identifying potential fault risks in advance.

[0041] Historical data sequence is The predicted output of the Long Short-Term Memory network model is ,when When the error exceeds the normal range, a fault warning signal will be issued.

[0042] Example 3 Passenger Service Module: This module uses communication devices to collect customer service requests and robots to provide services; it also builds an optimization model to generate behavioral instructions for each robot, with the goal of minimizing the total service waiting time.

[0043] In this embodiment, the communication device for collecting customer service requests is AR glasses.

[0044] 1. A Spatial-Temporal Graph Convolutional Network (ST-GCN) is used for demand prediction, combined with a greedy algorithm for resource allocation, to optimize the task queue of the service robot.

[0045] Service robot tasks include: food delivery, guidance, and cargo hold ventilation.

[0046] Service robot swarm: Employs UWB indoor positioning technology, achieving a positioning accuracy of up to 10cm; The ORCA algorithm is used to achieve multi-robot cooperative obstacle avoidance. By calculating the relative speed and distance between each robot and other obstacles, a safe obstacle avoidance speed is generated, enabling the robots to avoid collisions and successfully complete the task.

[0047] With the goal of minimizing the total service waiting time, an optimization model is constructed by combining physical constraints. The optimization model is solved by a greedy algorithm to generate robot behavior instructions.

[0048] Total service wait time:

[0049] in, This indicates the total number of service requests. Indicates the first The arrival time of each request, This indicates that the robot has begun processing the first... The time of each request , No. The waiting time for each request. Total service wait time.

[0050] Robot single-task constraint: A robot can only perform one task at any given time; Task completion constraint: Each service request must be assigned to one and only one robot; Physical sequence constraint: The robot must execute tasks in the order they appear in its task queue, and the start time of a later task must not be earlier than the end time of the previous task. Movement time constraint: The robot is for the request The service start time shall not be earlier than the sum of the request arrival time and the time required for the robot to move to the requested location.

[0051] All service robot task queues Initialize to empty; Record the next idle time of all robots. ; According to the requested predicted arrival time Sequential processing, for the current request Iterate through each robot : Calculate the estimated start time: ;in, For robots Move to request Time required at the location; calculate estimated waiting time ; Request Assigned to enable The smallest robot ,in ; Request Join the robot Task queue At the end, and update the robot. Next free time: , It is a service request. The time required for the event itself.

[0052] The behavioral instructions are: Task allocation matrix A binary matrix that defines which robot serves which request; Task queue collection : ,in It is an ordered list containing the assignments to the robots. All tasks are arranged in the order of execution; Each robot operates according to its own unique and ordered system. The task queue executes tasks sequentially.

[0053] In this embodiment, the robot 01 task queue is: [deliver meals to cabin A12, guide passengers to the entertainment center, deliver meals to cabin B05].

[0054] Robot 02 Task Queue: [Guide passengers to the service desk and deliver meals to cabin C10].

[0055] The UWB positioning system acquires passengers' location information in real time and transmits it to AR glasses. The AR glasses then display relevant service information and navigation guidance based on the passengers' location and needs.

[0056] When passengers arrive at the restaurant, the AR device can display the restaurant's menu, the current number of diners, and the estimated wait time. AR glasses can also provide personalized service recommendations.

[0057] A deep learning-based recommendation model generates a recommendation set based on passengers' historical service requests.

[0058] In terms of navigation, AR glasses use path planning algorithms to plan the optimal path based on the passenger's location and the target location obtained from the UWB positioning system, and then overlay the path as a virtual arrow in the passenger's real-time field of vision to achieve AR navigation.

[0059] AR glasses can also trigger an emergency response mechanism in a timely manner when wearable devices detect SOS signals sent by passengers or detect abnormal situations such as fires or drownings.

[0060] Example 4 The safety evacuation module integrates fire sensors, water immersion sensors, and passenger location data to dynamically generate the optimal evacuation route.

[0061] Fire sensors monitor parameters such as temperature and smoke concentration inside the ship's cabin in real time, while water immersion sensors monitor water ingress into the cabin. Temperature and smoke concentration thresholds for the fire sensors are set; a fire is detected when the sensor reading exceeds the set threshold, and an emergency event is detected when the smoke concentration exceeds a preset threshold.

[0062] Algorithm A is a heuristic search algorithm, often used for path planning and graph search problems.

[0063] By evaluating the function To select the optimal path: From the starting point to the node The actual cost (cost of the path already taken); From node The estimated cost to reach the destination (heuristic function, such as Euclidean distance, Manhattan distance); :node The total estimated cost; Algorithm A will prioritize expansion. Find the smallest node to improve search efficiency while ensuring the shortest path is found.

[0064] In this embodiment, the impact of emergency events, such as the direction of fire spread and the expansion of the flooded area, are incorporated into the evaluation function.

[0065] The evaluation function is defined as follows:

[0066] in, From the starting point to the node The actual cost; For the node The estimated cost to reach the destination; For nodes The total estimated cost; This is a weighting coefficient, dynamically adjusted according to the urgency of the emergency; when the emergency is more urgent, it is increased. The value of makes the algorithm more inclined to choose the shorter path; when the emergency is relatively mild, the value of is reduced. The value is determined by comprehensively considering the safety and distance of the path.

[0067] Dynamic weights adjust the weights of the heuristic function based on the urgency of emergencies such as fires and flooding, achieving a balance between safety and efficiency. By combining data from fire sensors, flood sensors, and passenger location data, the feasibility and safety of evacuation routes are updated in real time. In the event of fire or flooding, dangerous routes are dynamically closed and alternative routes are activated to ensure the real-time optimality of evacuation routes. Safety factors are incorporated into the cost function to prevent passengers from being guided through dangerous areas.

[0068] Example 5 System stability and performance optimization: 1. By collaborating edge computing with cloud computing, the load on the cloud can be reduced and the system response speed improved. Edge computing nodes can be deployed on ships to perform some real-time-critical computing tasks (such as preliminary processing of sensor data and local decision-making), thereby reducing the latency of data transmission to the cloud.

[0069] Let the computing power of the edge computing node be... The computing power of the cloud is The total number of computational tasks to be processed is T; the tasks are classified according to their real-time requirements, with tasks with high real-time requirements accounting for a certain percentage. The workload allocated to the edge computing nodes is The amount of tasks allocated to the cloud is .

[0070] Edge computing nodes and the cloud communicate via high-speed networks to achieve data synchronization and sharing. When an edge computing node cannot complete certain complex computing tasks independently, the relevant data is transmitted to the cloud for further processing, and the processing results are returned to the edge computing node. By optimizing network bandwidth and data transmission protocols, data transmission time is reduced, and the overall performance of the system is improved.

[0071] 2. To ensure system stability, redundant design and fault recovery mechanisms are employed. On the hardware side, critical equipment (such as sensors, computing nodes, and communication equipment) are redundantly configured. When a device fails, the system automatically switches to a backup device to ensure normal system operation. Assume the reliability of the equipment is... After adopting redundant design, the system reliability is It can be calculated based on the redundancy method. For example, for a parallel redundant system, , This refers to the number of redundant devices.

[0072] On the software side, a distributed architecture and fault-tolerant mechanisms are adopted, distributing system functions across multiple nodes to prevent single-point failures from causing system-wide crashes. Simultaneously, real-time monitoring and fault diagnosis are performed on the system. When a fault is detected, appropriate recovery measures are taken promptly, such as restarting the faulty node and reallocating tasks. Assume the system's failure rate is... The fault recovery time is Then the availability A of the system can be expressed as: A = MTBF / (MTBF + MTTR) Where MTBF is the mean time between failures, is MTTR stands for Mean Time to Repair and Time to Recovery. Related. By optimizing system design and fault recovery strategies, system availability can be improved, ensuring long-term stable operation.

[0073] Example 6 To verify the actual effectiveness of the system in passenger services, multi-scenario comparative experiments were conducted, covering hardware deployment, algorithm performance, and user experience evaluation. Experimental data was collected based on real ship environments and simulated scenarios to ensure the reliability of the results. Experimental configuration: Passenger service module: Hardware configuration: UWB positioning base station (Decawave DWM3000), Microsoft, HoloLens2AR device; Software: ST-GCN spatiotemporal graph network, Dijkstra's algorithm; Dataset: Anonymized passenger trajectory dataset, 6 months of operational logs.

[0074] Experimental design and evaluation metrics: Compared to commercial systems that use Wi-Fi positioning with an accuracy >1m, this method uses UWB+AR contactless services to improve positioning accuracy to 10cm, with a response latency <200ms, reducing physical interaction requirements by 90%. It also adds dynamic evacuation route planning by fusing fire or flooding sensor data with passenger location data. Figure 2 The system monitors the fire situation and pedestrian congestion data, closing dangerous fire routes in real time while simultaneously activating safer alternative routes (such as A→D→Exit). When the fire is not spreading, the system offers multiple routes; when the fire spreads to Zone B, the system monitors the fire to prevent its continued spread and closes the route (A→B→Exit) to prevent poisoning or burns in that area, while simultaneously activating the backup route (A→D→Exit); in the later stages of the fire, due to the fire and pedestrian congestion exceeding 80%, the route (A→C→Exit) becomes severely congested. The system then prompts and suggests that pedestrians behind the congestion abandon the route (A→C→Exit) and activate the only reliable route (A→D→Exit) to prevent stampedes. Furthermore, dynamic evacuation route planning is added by integrating fire or flooding sensor data with passenger location data, reducing route update latency to <200ms (compared to 300ms for commercial systems). The evacuation time for commercial systems is 12.6 minutes, while our system reduces it by 35% to 8.2 minutes. Additionally, ST-GCN demand forecasting is added, using a spatiotemporal graph model to predict service hotspots. Compared to the traditional LSTM model (72%), our system achieves an accuracy of 89%, and supports real-time monitoring of peak-hour passenger flow. The performance of the contactless service is shown in Table 1. Table 1

[0075] Statistical significance verification: t-test analysis (95% confidence level) showed that the temperature fluctuation of the cargo in our system was p=0.003 compared to the traditional system, and the service response time was p=0.001 compared to Smith et al. 2021. Both experimental results passed the significance test (p<0.05). Multimodal data fusion: cargo recognition accuracy improved by 12%, relying on complementary features of RGB and point cloud; edge computing optimization: service response latency reduced to milliseconds, reducing cloud dependence; dynamic strategy adaptation: digital twin model improved temperature control efficiency by 75%.

[0076] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multimodal autonomous perception and integrated management system for intelligent ships, characterized in that, include: The system includes a route planning module, a fault detection module, a passenger service module, and a safety evacuation module. The path planning module performs path planning based on real-time collected ship environmental data; The fault detection module performs real-time fault detection based on ship environmental data. The passenger service module includes: Communication devices that collect customer service requests and robots that provide services; An optimization model is constructed with the goal of minimizing the total service wait time to generate behavioral instructions for each robot. The safety evacuation module integrates fire sensors, water immersion sensors, and passenger location data to dynamically generate the optimal evacuation route.

2. The system according to claim 1, characterized in that, The route planning based on real-time collected ship environmental data includes: The system collects real-time data on the ship's surrounding environment using sensors, including the location of static obstacles, the trajectory and speed of dynamic obstacles, typhoon paths, and meteorological data. Multiple initial feasible paths connecting the starting point and the target point are generated based on the A* algorithm, where the path nodes satisfy the minimum turning radius and speed constraints of the ship. Based on the predicted trajectory of dynamic obstacles, spatiotemporal conflict detection is performed on the initial path, and the path is corrected in conjunction with the collision avoidance priority in the COLREGs rules. The correction includes: When encountering a vessel approaching from the starboard side, turn to the starboard to avoid it; when encountering a vessel crossing from the port side, maintain course and reduce speed. When the direction of typhoon movement blocks the shipping channel, the detour route to the left of the typhoon's path is selected first. If this is not feasible, an anchorage waiting route is generated. For each candidate path, a multi-objective evaluation is performed, including: Calculate the efficiency score and the safety score of minimum obstacle distance under the fuel consumption model, and generate the comprehensive optimal path through a weighted function; Output the optimal path and dynamically replan during the voyage by periodically updating the data.

3. The system according to claim 1, characterized in that, Total service wait time: in, This indicates the total number of service requests. Indicates the first The arrival time of each request, This indicates that the robot has begun processing the first... The time of each request , No. The waiting time for each request. Total service wait time.

4. The system according to claim 3, characterized in that, The optimization model also includes constraints: Robot single-task constraint: A robot can only perform one task at any given time; Task completion constraint: Each service request must be assigned to one and only one robot; Physical sequence constraint: The robot must execute tasks in the order they appear in its task queue, and the start time of a later task must not be earlier than the end time of the previous task. Movement time constraint: The robot is for the request The service start time shall not be earlier than the sum of the request arrival time and the time required for the robot to move to the requested location.

5. The system according to claim 3, characterized in that, Solve the optimization model to generate behavioral instructions for each robot: All service robot task queues Initialize to empty; Record the next idle time of all robots. ; According to the requested predicted arrival time Sequential processing, for the current request Iterate through each robot : Calculate the estimated start time: ;in, For robots Move to request Time required at the location; calculate estimated waiting time ; Request Assigned to enable The smallest robot ,in ; Request Join the robot Task queue At the end, and update the robot. Next free time: , It is a service request. The time required for the event itself.

6. The system according to claim 3, characterized in that, The behavioral instruction is: Task allocation matrix A binary matrix that defines which robot serves which request; Task queue collection : ,in It is an ordered list containing the assignments to the robots. All tasks are arranged in the order of execution; Each robot operates according to its own unique and ordered system. The task queue executes tasks sequentially.

7. The system according to claim 1, characterized in that, The system integrates fire sensor, water immersion sensor, and passenger location data to dynamically generate the optimal evacuation route, including: The evaluation function is: in, From the starting point to the node The actual cost; For the node The estimated cost to reach the destination; For nodes The total estimated cost; This is a weighting coefficient that is dynamically adjusted based on the urgency of the emergency; when the emergency is more urgent, it is increased. The value of makes the algorithm more inclined to choose the shorter path; when the emergency is relatively mild, the value of is reduced. The value is determined by comprehensively considering the safety and distance of the path.

8. The system according to claim 1, characterized in that, The real-time fault detection based on ship environmental data includes: The vibration sensor collects the vibration signal of the engine in real time. When an abnormal peak value appears within the preset frequency range, it is judged to be a bearing wear fault. Temperature sensors monitor temperature changes in the engine room. When the temperature sensor reading exceeds a preset temperature threshold, it is determined that the engine room is overheating. The pressure sensor monitors the pressure in the oil pipeline, and when the pressure sensor reading is lower than the preset pressure threshold, it is determined to be a fuel leak.

Images