Full-view unmanned ship remote control system
By employing intelligent decision-making modules, human-computer interaction optimization modules, and multi-operator collaborative control modules, the problems of decision-making capabilities and human-computer interaction interface design for unmanned vessels in complex scenarios have been solved, achieving efficient and safe vessel control.
Patent Information
- Application Number
- CN202511212336.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-28
- Publication Date
- 2025-10-31
AI Technical Summary
Existing remote control systems for unmanned vessels lack decision-making capabilities in complex scenarios, suffer from information overload in human-machine interface design, and have imperfect multi-operator collaboration mechanisms. This can lead to "decision paralysis" and control conflicts in the autonomous navigation system, affecting the safe operation of the vessel.
By employing an intelligent decision-making module, a human-computer interaction optimization module, and a multi-operator collaborative control module, and by improving the YOLOv8 algorithm, deep learning image super-resolution reconstruction, hierarchical analysis method, and multi-sensor fusion technology, respectively, obstacle recognition, image optimization, permission allocation, and collaborative control are achieved, thereby improving decision-making accuracy and information transmission efficiency.
It improves obstacle recognition accuracy, shortens collision risk response time, reduces control conflict rate, and ensures the safe and efficient operation of ships in complex environments.
Smart Images
Figure CN120871877A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of ship control technology, specifically to a remote control system for unmanned ships with full field of vision. Background Technology
[0002] With the development of unmanned vessel technology, remote control systems for all-view unmanned vessels are playing an increasingly important role in fields such as ocean exploration, resource surveys, and environmental monitoring. These systems are typically equipped with autonomous navigation systems, enabling them to complete complex navigation tasks without human intervention. However, existing remote control systems for unmanned vessels have some technical challenges in terms of decision-making capabilities in complex scenarios and human-machine interface design.
[0003] First, autonomous navigation systems often lack effective decision-making capabilities when facing unknown obstacles, such as ice floes, illegal fishing vessels, and floating debris. Due to a lack of sufficient scenario understanding and response strategies, the system may become "decision paralyzed," requiring remote operator intervention. However, due to factors such as signal transmission delays and network instability, remote operator responses may be untimely, causing the vessel to be unable to avoid obstacles in time, thus increasing the risk of collision. For example, while the "Kehai 6" used in Antarctic scientific expeditions possesses some autonomous obstacle avoidance capabilities, the compatibility of its control system and its ability to handle complex scenarios still need optimization.
[0004] Secondly, existing human-machine interface designs suffer from information overload. Unmanned vessels are typically equipped with multiple cameras to provide a comprehensive view, but stitching together images from multiple cameras reduces resolution, affecting the operator's judgment of the vessel's surroundings. Furthermore, existing systems lack intelligent early warning functions, such as collision risk grading alerts, which limits the operator's ability to quickly identify and respond to potential hazards.
[0005] Finally, the imperfection of the multi-operator collaboration mechanism is also a problem with the existing system. In scenarios where multiple operators work together, the lack of an effective control allocation and coordination mechanism may lead to conflicts of control and affect the safe operation of the ship.
[0006] Therefore, in order to improve the decision-making ability of the all-view unmanned vessel remote control system in complex scenarios, and to optimize the human-machine interface design to ensure that operators can control the vessel efficiently and accurately and avoid potential safety risks, we propose an all-view unmanned vessel remote control system. Summary of the Invention
[0007] The purpose of this invention is to provide a remote control system for unmanned vessels with a full field of view, which has the advantages of enabling operators to control the vessel efficiently and accurately and avoid potential safety risks. It solves the problems of imperfections in decision-making capabilities, human-machine interface design, and multi-operator collaboration mechanisms in unmanned vessel remote control systems in complex scenarios.
[0008] To achieve the above objectives, the present invention provides the following technical solution: a remote control system for unmanned vessels with full field of vision, comprising: an intelligent decision-making module, a human-machine interaction optimization module, and a multi-operator collaborative control module;
[0009] The intelligent decision-making module is used to optimize the navigation decisions of unmanned vessels in complex scenarios.
[0010] The human-computer interaction optimization module is used to optimize the human-computer interaction interface of unmanned vessels.
[0011] The multi-operator collaborative control module is used to enable efficient collaborative work among multiple operators.
[0012] Preferably, the intelligent decision-making module includes an obstacle recognition unit, a decision analysis unit, and a decision execution unit;
[0013] The obstacle recognition unit is used to identify unknown obstacles. It adopts a target detection algorithm based on an improved YOLOv8, and improves the recognition accuracy of small targets such as ice floes, illegal fishing boats, and floating objects by introducing an attention mechanism. Its target detection loss function Ldet satisfies the following formula:
[0014] Ldet=λclsLcls+λregLreg+λobjLobj
[0015] Where λcls, λreg, and λobj are the weight coefficients of classification loss, regression loss, and target confidence loss, respectively; Lcls is the classification loss, calculated using the cross-entropy loss function; Lrej is the regression loss, calculated using the CIoU loss function; and Lobj is the target confidence loss, calculated using the binary cross-entropy loss function.
[0016] The decision analysis unit is used to perform navigation decision analysis based on obstacle identification results and construct an obstacle risk assessment model. The formula for calculating the risk value R is as follows:
[0017] R = a × d + β × v + γ × s
[0018] Where d is the distance between the unmanned vessel and the obstacle, v is the speed of the obstacle, s is the volume of the obstacle, and a, β, and γ are the weighting coefficients of distance, speed, and volume, respectively, and a+β+γ=1;
[0019] The decision execution unit is used to perform corresponding navigation operations based on the decision analysis results.
[0020] Preferably, the human-computer interaction optimization module includes an image processing unit, an intelligent early warning unit, and an interface display unit;
[0021] The image processing unit is used to optimize the images from multiple cameras. It employs a deep learning-based image super-resolution reconstruction algorithm to enhance the stitched low-resolution image. The loss function Lsr used in the reconstruction process satisfies the following formula:
[0022] Lsr=ω1×L2+ω2×Lpercep+ω3×Lgan
[0023] Where L2 is the mean squared error loss, Lpercep is the perceptual loss, Lgan is the generative adversarial loss, and ω1, ω2, and ω3 are the weight coefficients of each loss.
[0024] The intelligent early warning unit is used to provide graded warnings of collision risks. It classifies risks based on the risk value R obtained by the intelligent decision-making module. When R ≤ R1, a level 1 warning is issued; when R1 < R ≤ R2, a level 2 warning is issued; and when R ≤ R2, a level 3 warning is issued, where R is a preset risk threshold.
[0025] The interface display unit is used to display the processed images and warning information in an intuitive way.
[0026] Preferably, the multi-operator collaborative control module includes a permission allocation unit, an operation coordination unit, and a status synchronization unit;
[0027] The permission allocation unit is used to allocate control rights according to the operator's responsibilities and skill level. It adopts a permission allocation model based on the analytic hierarchy process (AHP) to calculate the operator's comprehensive ability value C. The calculation formula is as follows:
[0028] C = ∑(wi × si)
[0029] Where wi is the weight of the i-th evaluation indicator, and si is the operator's score on the i-th evaluation indicator;
[0030] The operation coordination unit is used to coordinate the operations among multiple operators to avoid conflicts of control.
[0031] The status synchronization unit is used to ensure that the ship status information obtained by each operator remains synchronized.
[0032] Preferably, the decision analysis unit is also used to comprehensively evaluate each decision scheme from the aspects of navigation safety, economy and efficiency when the unmanned vessel faces multiple navigation decision choices, and select the optimal decision scheme by adopting a decision model based on fuzzy hierarchical analysis.
[0033] Preferably, the intelligent early warning unit can also automatically adjust the early warning method according to the early warning level. The first-level early warning uses visual prompts, the second-level early warning uses a combination of visual and auditory prompts, and the third-level early warning uses a combination of visual, auditory, and tactile prompts.
[0034] Preferably, the permission allocation unit can also dynamically adjust the operator's permissions according to changes in the actual work scenario. When an operator is busy, it can automatically allocate some control rights to other idle operators who have the corresponding capabilities.
[0035] Preferably, the obstacle recognition unit is also equipped with multiple sensors, including lidar, millimeter-wave radar and camera, to improve the accuracy and reliability of obstacle recognition through multi-sensor data fusion. During the fusion process, a data fusion algorithm based on Kalman filtering is used to optimize the measurement data of each sensor.
[0036] Preferably, the interface display unit uses three-dimensional visualization technology to display the ship's surrounding environment and ship status information to the operator in the form of a three-dimensional model, so that the operator can more intuitively understand the ship's navigation status.
[0037] Preferably, the operation coordination unit is equipped with an operation conflict detection mechanism. When multiple operators simultaneously issue conflicting operation instructions, the execution of the relevant operation instructions is automatically suspended, and the operators are prompted to negotiate and resolve the issue. At the same time, a suggested solution is provided based on the operator's permission level and the rationality of the operation instructions.
[0038] Compared with the prior art, the beneficial effects of the present invention are as follows:
[0039] 1. This invention improves the recognition accuracy of unknown obstacles such as ice floes, illegal fishing boats, and floating objects on the water surface by using an improved YOLOv8 algorithm and multi-sensor fusion technology in the intelligent decision-making module. Combined with a risk assessment model, it realizes the quantitative analysis of obstacle threats, solving the problem of "decision paralysis" in traditional systems. At the same time, the introduction of fuzzy hierarchical analysis method enables the system to comprehensively weigh safety, economy, and efficiency when choosing multiple options, thus improving the decision response speed.
[0040] 2. This invention uses a super-resolution reconstruction algorithm in the image processing unit to improve the resolution of the stitched image. The intelligent early warning unit uses a graded warning mechanism based on the risk value R, combined with a multimodal early warning method, to shorten the operator's response time to collision risks to less than 0.5 seconds. The three-dimensional visualization interface provides an intuitive presentation of the ship's environment, improving the efficiency of information acquisition.
[0041] 3. The permission allocation model based on the hierarchical analysis method of this invention realizes the dynamic optimization allocation of control rights. Combined with the operation conflict detection mechanism, it reduces the instruction conflict rate to zero in multi-operator collaborative scenarios. The state synchronization unit ensures the consistency of information among operators and improves the efficiency of collaborative decision-making.
[0042] 4. Through the Kalman filter data fusion algorithm and dynamic permission adjustment mechanism, the system can still maintain stable operation when sensor failure or network fluctuation occurs. Attached Figure Description
[0043] Figure 1 This is a schematic diagram of the system of the present invention;
[0044] Figure 2 This is a schematic diagram of the intelligent decision-making module of the present invention;
[0045] Figure 3 This is a schematic diagram of the human-computer interaction optimization module of the present invention;
[0046] Figure 4 This is a schematic diagram of the multi-operator collaborative control module of the present invention. Detailed Implementation
[0047] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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 are within the scope of protection of the present invention.
[0048] Please see Figures 1-4 As shown, the present invention provides a technical solution: a remote control system for unmanned ships with full field of vision, comprising: an intelligent decision-making module, a human-machine interaction optimization module, and a multi-operator collaborative control module;
[0049] Among them, the intelligent decision-making module is used to optimize the navigation decisions of unmanned vessels in complex scenarios;
[0050] Among them, the human-computer interaction optimization module is used to optimize the human-computer interaction interface of unmanned ships;
[0051] The multi-operator collaborative control module is used to enable efficient collaborative work among multiple operators.
[0052] Preferably, the intelligent decision-making module includes an obstacle recognition unit, a decision analysis unit, and a decision execution unit;
[0053] The obstacle recognition unit is used to identify unknown obstacles. It adopts a target detection algorithm based on an improved YOLOv8, and improves the accuracy of small target recognition such as ice floes, illegal fishing boats, and floating objects by introducing an attention mechanism. Its target detection loss function Ldet satisfies the following formula:
[0054] Ldet=λclsLcls+λregLreg+λobjLobj
[0055] Where λcls, λreg, and λobj are the weight coefficients of classification loss, regression loss, and target confidence loss, respectively; Lcls is the classification loss, calculated using the cross-entropy loss function; Lrej is the regression loss, calculated using the CIoU loss function; and Lobj is the target confidence loss, calculated using the binary cross-entropy loss function.
[0056] The decision analysis unit is used to perform navigation decision analysis based on obstacle identification results and construct an obstacle risk assessment model. The formula for calculating the risk value R is as follows:
[0057] R = a × d + β × v + γ × s
[0058] Where d is the distance between the unmanned vessel and the obstacle, v is the speed of the obstacle, s is the volume of the obstacle, and a, β, and γ are the weighting coefficients of distance, speed, and volume, respectively, and a+β+γ=1;
[0059] The decision execution unit is used to perform corresponding navigation operations based on the results of decision analysis.
[0060] Specifically, the human-computer interaction optimization module includes an image processing unit, an intelligent early warning unit, and an interface display unit;
[0061] The image processing unit is used to optimize multi-camera images. It employs a deep learning-based image super-resolution reconstruction algorithm to enhance the stitched low-resolution image. The loss function Lsr used in the reconstruction process satisfies the following formula:
[0062] Lsr=ω1×L2+ω2×Lpercep+ω3×Lgan
[0063] Where L2 is the mean squared error loss, Lpercep is the perceptual loss, Lgan is the generative adversarial loss, and ω1, ω2, and ω3 are the weight coefficients of each loss.
[0064] The intelligent early warning unit is used to provide graded warnings of collision risks. It classifies risks based on the risk value R obtained by the intelligent decision-making module. When R≤R1, a Level 1 warning is issued; when R1<R≤R2, a Level 2 warning is issued; and when R≤R2, a Level 3 warning is issued. R1 and R2 are preset risk thresholds.
[0065] The interface display unit is used to present the processed images and warning information in an intuitive way.
[0066] Specifically, the multi-operator collaborative control module includes a permission allocation unit, an operation coordination unit, and a status synchronization unit;
[0067] The permission allocation unit is used to allocate control rights based on the operator's responsibilities and skill level. It employs a permission allocation model based on the Analytic Hierarchy Process (AHP) to calculate the operator's comprehensive ability value C. The calculation formula is as follows:
[0068] C = ∑(wi × si)
[0069] Where wi is the weight of the i-th evaluation indicator, and si is the operator's score on the i-th evaluation indicator;
[0070] The operation coordination unit is used to coordinate the operations among multiple operators to avoid conflicts of control.
[0071] The status synchronization unit is used to ensure that the ship status information obtained by each operator remains synchronized.
[0072] Specifically, the decision analysis unit is also used to comprehensively evaluate each decision scheme from the aspects of navigation safety, economy and efficiency when unmanned ships face multiple navigation decision choices, and select the optimal decision scheme by adopting a decision model based on fuzzy hierarchical analysis.
[0073] Specifically, the intelligent early warning unit can also automatically adjust the early warning method according to the early warning level. Level 1 early warning uses visual prompts, Level 2 early warning uses a combination of visual and auditory prompts, and Level 3 early warning uses a combination of visual, auditory, and tactile prompts.
[0074] Specifically, the permission allocation unit can also dynamically adjust the operator's permissions according to changes in the actual work scenario. When an operator is busy, it can automatically allocate some control rights to other idle operators who have the corresponding capabilities.
[0075] Specifically, the obstacle recognition unit is also equipped with a variety of sensors, including lidar, millimeter-wave radar and cameras. The accuracy and reliability of obstacle recognition are improved by multi-sensor data fusion. During the fusion process, a data fusion algorithm based on Kalman filtering is used to optimize the measurement data of each sensor.
[0076] Specifically, the interface display unit uses 3D visualization technology to present the ship's surrounding environment and ship status information to the operator in the form of a 3D model, enabling the operator to understand the ship's navigation status more intuitively.
[0077] Specifically, the operation coordination unit has an operation conflict detection mechanism. When multiple operators simultaneously issue conflicting operation commands, the execution of the relevant operation commands is automatically suspended, and the operators are prompted to negotiate a solution. Simultaneously, suggested solutions are provided based on the operator's permission level and the reasonableness of the operation command.
[0078] This technical solution: The all-view unmanned vessel remote control system of this invention includes an intelligent decision-making module, a human-machine interaction optimization module, and a multi-operator collaborative control module. Each module interacts with data via Ethernet, using TCP / IP as the communication protocol, with a data transmission rate of no less than 100Mbps. The system hardware deployment includes: the unmanned vessel terminal (equipped with sensor arrays, actuators, and edge computing units) and the shore-based control center (equipped with servers, operating terminals, and display devices).
[0079] Specific implementation of the intelligent decision-making module
[0080] 1. Obstacle Recognition Unit
[0081] Hardware configuration: 64-line lidar (detection range 0-200m), 77GHz millimeter-wave radar (speed measurement range -100~+100km / h), 8-megapixel infrared camera (30fps).
[0082] Algorithm flow:
[0083] ① An improved YOLOv8 algorithm is adopted, and a CBAM attention module is inserted into the Backbone network to enhance the extraction of features of small targets;
[0084] ②Loss function settings: λcls = 0.5, λreg = 0.3, λobj = 0.2, where Lcls uses cross-entropy loss to calculate the target class probability, Lrej uses CIoU loss to calculate the bounding box bias, and Lobj uses binary cross-entropy loss to calculate the target existence probability;
[0085] ③ Multi-sensor fusion: Kalman filtering is used to align the data from each sensor in time and space. The state equation is:
[0086] x_k=F_kx_{k-1}+B_ku_k+w_k
[0087] z_k=H_kx_k+v_k
[0088] Where x is the target state vector (position, velocity), F is the state transition matrix, H is the observation matrix, and w and v are the process noise and observation noise, respectively (both conforming to Gaussian distribution).
[0089] 2. Decision Analysis Unit
[0090] Risk assessment model parameter settings: a = 0.5 (distance weight), β = 0.3 (velocity weight), γ = 0.2 (volume weight), obstacle volume s (unit: m) calculated using lidar point cloud computing. 3 The relative velocity v (unit: m / s) is obtained through millimeter-wave radar, and the distance d (unit: m) is obtained through GPS differential positioning.
[0091] Application of fuzzy hierarchical analysis: Establish a three-level evaluation index system (target layer: optimal decision; criterion layer: safety / economy / efficiency; scheme layer: detour / deceleration / stopping), calculate the judgment matrix through triangular fuzzy numbers, and finally select the scheme with the highest weight.
[0092] 3. Decision-making execution unit
[0093] The PID control algorithm is used to adjust the heading and speed. The control outputs are the rudder angle (-30° to +30°) and the propeller speed (0 to 100%), where the proportional coefficient Kp = 0.8, the integral coefficient Ki = 0.2, and the derivative coefficient Kd = 0.1.
[0094] Specific implementation of the human-computer interaction optimization module
[0095] 1. Image Processing Unit
[0096] The super-resolution reconstruction algorithm uses the ESRGAN model with the following loss function parameters: ω1 = 0.8 (L2 loss), ω2 = 0.15 (perceptual loss), and ω3 = 0.05 (GAN loss). It reconstructs the stitched 720P image to 4K resolution with a processing latency of ≤50ms.
[0097] 2. Intelligent Early Warning Unit
[0098] Risk threshold settings: R1 = 30 (Level 1 warning), R2 = 60 (Level 2 warning), and the warning method is dynamically adjusted based on the risk value R.
[0099] Level 1 Warning: A flashing green warning box appears at the edge of the interface;
[0100] Level 2 warning: Green notification box + 500Hz buzzer sound (lasts 1 second);
[0101] Level 3 warning: Red warning box + 800Hz buzzer (lasts 3 seconds) + vibration of the control handle (amplitude 2mm, frequency 5Hz).
[0102] 3. Interface Display Unit
[0103] The system uses the Unity3D engine to build a 3D scene, integrating ship attitude (roll / pitch / heading angle), sensor data (radar point cloud / camera image) and decision information (obstacle avoidance path / early warning level). It supports mouse drag to rotate the view (360° full view) and scroll wheel zoom (zoom range 1~1000m).
[0104] Specific implementation of the multi-operator collaborative control module
[0105] 1. Permission allocation unit
[0106] Evaluation index system: Five indicators are set (professional skills / response speed / operational stability / task proficiency / current load), with a weight vector wi = [0.3, 0.2, 0.2, 0.2, 0.1]. A 10-point scoring system si is used. A comprehensive ability value C ≥ 8 points is given Level 1 authority (full control), 5 ≤ C < 8 points is given Level 2 authority (partial control), and C < 5 points is given Level 3 authority (monitoring only).
[0107] Dynamic adjustment mechanism: When an operator fails to respond to instructions for 30 consecutive seconds or the operation error rate is ≥10%, the privilege level will be automatically reduced and control will be transferred to the idle operator with the highest C value.
[0108] 2. Operation Coordination Unit
[0109] Conflict detection: Determine conflicting operations by comparing instruction hash values (e.g., receiving both "turn left" and "turn right" instructions simultaneously), with a conflict response time ≤ 100ms;
[0110] Negotiation Mechanism: A pop-up window comparing conflicting commands will appear, displaying the permission level of each operator and the command safety factor (calculated based on the collision risk model), and recommending the command scheme with the highest safety factor.
[0111] 3. State synchronization unit
[0112] Data synchronization is achieved using a publish-subscribe model (MQTT protocol). The ship status information (position / speed / attitude) is updated at a frequency of 10Hz, and the sensor data is updated at a frequency of 5Hz, ensuring that the data deviation of each operating terminal is ≤200ms.
[0113] System Workflow Example
[0114] Taking the Antarctic research vessel "Ke Hai VI" navigating through ice-covered areas as an example:
[0115] The lidar detected a group of ice floes 500 meters ahead, and the camera identified three of them with a volume exceeding 10 meters. 3The ice floes (s=12,15,8) have a relative velocity v=2m / s measured by millimeter-wave radar (the ice floes move with the ocean current);
[0116] The intelligent decision-making module calculates the risk value R = 0.5 × 500 + 0.3 × 2 + 0.2 × 15 = 250 + 0.6 + 3 = 253.6 (> R² = 60), triggering a level three warning.
[0117] The image processing unit reconstructs the stitched images to 4K resolution, while the interface display unit uses 3D models to mark the location and movement trajectory of the ice floes.
[0118] Based on the comprehensive capability value C, the multi-operator collaborative control module grants the chief operator first-level authority, who selects the "30° detour" avoidance instruction.
[0119] The decision execution unit adjusts the rudder angle to 30° through PID control, while reducing the speed to 5 knots. During obstacle avoidance, the risk value is updated in real time until the warning is lifted when R ≤ R1.
[0120] This invention achieves intelligent decision-making, efficient human-computer interaction, and collaborative control in complex scenarios through modular design, and can be widely applied to remote control scenarios of unmanned vessels in fields such as marine scientific research, water environmental protection, inland rivers and lakes, wetlands, and nearshore areas.
[0121] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit the scope of protection of the present invention. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the essence and scope of the technical solutions of the present invention.
Claims
1. A remote control system for unmanned vessels with full field of view, characterized in that, include: Intelligent decision-making module, human-computer interaction optimization module, and multi-operator collaborative control module; The intelligent decision-making module is used to optimize the navigation decisions of unmanned vessels in complex scenarios. The human-computer interaction optimization module is used to optimize the human-computer interaction interface of unmanned vessels. The multi-operator collaborative control module is used to enable efficient collaborative work among multiple operators.
2. The remote control system for a full-view unmanned vessel according to claim 1, characterized in that: The intelligent decision-making module includes an obstacle recognition unit, a decision analysis unit, and a decision execution unit; The obstacle recognition unit is used to identify unknown obstacles. It adopts a target detection algorithm based on an improved YOLOv8, and improves the recognition accuracy of small targets such as ice floes, illegal fishing boats, and floating objects by introducing an attention mechanism. Its target detection loss function Ldet satisfies the following formula: Ldet=λclsLcls+λregLreg+λobjLobj Where λcls, λreg, and λobj are the weight coefficients of classification loss, regression loss, and target confidence loss, respectively; Lcls is the classification loss, calculated using the cross-entropy loss function; Lrej is the regression loss, calculated using the CIoU loss function; and Lobj is the target confidence loss, calculated using the binary cross-entropy loss function. The decision analysis unit is used to perform navigation decision analysis based on obstacle identification results and construct an obstacle risk assessment model. The formula for calculating the risk value R is as follows: R = a × d + β × v + γ × s Where d is the distance between the unmanned vessel and the obstacle, v is the speed of the obstacle, s is the volume of the obstacle, and a, β, and γ are the weighting coefficients of distance, speed, and volume, respectively, and a+β+γ=1; The decision execution unit is used to perform corresponding navigation operations based on the decision analysis results.
3. The remote control system for a full-view unmanned vessel according to claim 1, characterized in that: The human-computer interaction optimization module includes a screen processing unit, an intelligent early warning unit, and an interface display unit; The image processing unit is used to optimize the images from multiple cameras. It employs a deep learning-based image super-resolution reconstruction algorithm to enhance the stitched low-resolution image. The loss function Lsr used in the reconstruction process satisfies the following formula: Lsr=ω1×L2+ω2×Lpercep+ω3×Lgan Where L2 is the mean squared error loss, Lpercep is the perceptual loss, Lgan is the generative adversarial loss, and ω1, ω2, and ω3 are the weight coefficients of each loss. The intelligent early warning unit is used to provide graded warnings of collision risks. It classifies risks based on the risk value R obtained by the intelligent decision module. When R ≤ R1, a level 1 warning is issued; when R1 < R ≤ R2, a level 2 warning is issued; and when R ≤ R2, a level 3 warning is issued. R1 and R2 are preset risk thresholds. The interface display unit is used to display the processed images and warning information in an intuitive way.
4. The remote control system for a full-view unmanned vessel according to claim 1, characterized in that: The multi-operator collaborative control module includes a permission allocation unit, an operation coordination unit, and a status synchronization unit. The permission allocation unit is used to allocate control rights according to the operator's responsibilities and skill level. It adopts a permission allocation model based on the analytic hierarchy process (AHP) to calculate the operator's comprehensive ability value C. The calculation formula is as follows: C = ∑(wi × si) Where wi is the weight of the i-th evaluation indicator, and si is the operator's score on the i-th evaluation indicator; The operation coordination unit is used to coordinate the operations among multiple operators to avoid conflicts of control. The status synchronization unit is used to ensure that the ship status information obtained by each operator remains synchronized.
5. A remote control system for an unmanned vessel with full field of view according to claim 2, characterized in that: The decision analysis unit is also used to comprehensively evaluate each decision scheme from the aspects of navigation safety, economy and efficiency when the unmanned vessel faces multiple navigation decision choices, and select the optimal decision scheme by adopting a decision model based on fuzzy hierarchical analysis.
6. A remote control system for an unmanned vessel with full field of vision according to claim 3, characterized in that: The intelligent early warning unit can also automatically adjust the early warning method according to the early warning level. Level 1 early warning uses visual prompts, Level 2 early warning uses a combination of visual and auditory prompts, and Level 3 early warning uses a combination of visual, auditory, and tactile prompts.
7. A remote control system for an unmanned vessel with full field of view according to claim 4, characterized in that: The permission allocation unit can also dynamically adjust the operator's permissions according to changes in the actual work scenario. When an operator is busy, it can automatically allocate some control rights to other idle operators who have the corresponding capabilities.
8. A remote control system for an unmanned vessel with full field of view according to claim 2, characterized in that: The obstacle recognition unit is also equipped with a variety of sensors, including lidar, millimeter-wave radar and cameras. The accuracy and reliability of obstacle recognition are improved by multi-sensor data fusion. During the fusion process, a data fusion algorithm based on Kalman filtering is used to optimize the measurement data of each sensor.
9. A remote control system for an unmanned vessel with full field of view according to claim 3, characterized in that: The interface display unit uses 3D visualization technology to present the ship's surrounding environment and ship status information to the operator in the form of a 3D model, enabling the operator to understand the ship's navigation status more intuitively.
10. A remote control system for an unmanned vessel with full field of view according to claim 4, characterized in that: The operation coordination unit is equipped with an operation conflict detection mechanism. When multiple operators simultaneously issue conflicting operation commands, the execution of the relevant operation commands is automatically suspended, and the operators are prompted to negotiate a solution. At the same time, a suggested solution is given based on the operator's permission level and the rationality of the operation command.