Unmanned ship path optimization method based on visual detection

Abstract

The invention discloses an unmanned ship path optimization method based on visual detection, and belongs to the technical field of unmanned ships, and the method comprises the steps: receiving first image information collected by a visual sensor carried by an unmanned ship and navigation parameter information from the unmanned ship, and constructing an environment perception model; performing motion trend prediction and unmanned ship collision risk comprehensive evaluation on the dynamic target; constructing a multi-dimensional threat assessment function, carrying out quantitative sorting on all known collision risk sources, defining a threat assessment index for each risk source, and optimizing a multi-target conflict logic; and dynamically adjusting the course and speed of the unmanned ship based on the output threat assessment result. In the implementation process of the technical scheme, the surrounding environment is perceived in real time through visual detection, parameters such as obstacle types, distances, relative speeds and movement directions are input into a multi-dimensional threat assessment function, and dynamic weighted sorting of different risk sources is realized, so that navigation path optimization is performed on the unmanned ship.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned ships, in particular to an unmanned ship path optimization method based on visual detection. BACKGROUND

[0002] An unmanned ship is an intelligent ship that can autonomously navigate on the water surface without human intervention, and is widely used in environmental monitoring, water patrol, surveying and mapping, and other fields.

[0003] With the rapid development of artificial intelligence and computer vision technology, intelligentization has become one of the core directions of the development of unmanned ships. Visual detection technology provides key support for the autonomous navigation of unmanned ships in complex water environments, enabling them to identify obstacles, channel boundaries, and water surface dynamic targets in real time, thereby providing precise environmental perception capabilities for unmanned ships and providing precise path planning and obstacle avoidance decision-making basis.

[0004] However, there are still some problems in the existing visual navigation process of unmanned ships. For example, when multiple static or dynamic obstacles are detected, such as coexistence of dense floating objects and lateral high-speed vessels, existing methods usually use simple strategies such as nearest obstacle priority or fixed weight ordering, without considering other factors such as target type and motion threat level, resulting in rigid obstacle avoidance decisions and non-optimal navigation planning results, making it difficult to meet the real-time and safety requirements in complex and dynamic water environments.

[0005] Therefore, it is necessary to provide an unmanned ship path optimization method based on visual detection to solve the above problems.

[0006] It should be noted that the above information disclosed in this background section is only used to understand the background technology of the present application, and therefore, it can contain information that does not constitute prior art. SUMMARY

[0007] Based on the above problems existing in the prior art, the present application solves the problem of providing an unmanned ship path optimization method based on visual detection, which optimizes the navigation route of the unmanned ship to make the navigation process more accurate and effective.

[0008] The technical solution adopted by the present application to solve its technical problems is: an unmanned ship path optimization method based on visual detection, comprising: The data processing center receives first image information collected by the visual sensor carried by the unmanned ship, and pre-processes the first image information, and receives navigation parameter information from the unmanned ship, and calibrates the navigation parameter information to construct an environment perception model; Based on the data output by the aforementioned environment perception model, the motion trend of the dynamic target is predicted and the collision risk of the unmanned ship is comprehensively evaluated; A multi-dimensional threat assessment function is constructed to quantitatively sort all known collision risk sources, define a threat assessment index for each risk source, and optimize multi-objective conflict logic; Based on the output threat assessment result, the heading and speed of the unmanned ship are dynamically adjusted, and a smooth obstacle avoidance trajectory is generated in combination with environmental constraints and a dynamics model.

[0009] In the implementation process of the technical solutions of the present application, the surrounding environment is perceived in real time through visual detection, and parameters such as obstacle type, distance, relative speed, and motion direction are input into the multi-dimensional threat assessment function to realize dynamic weighted sorting of different risk sources, thereby optimizing the navigation path of the unmanned ship.

[0010] Further, the data processing center performs spatio-temporal alignment and feature-level fusion on the preprocessed image information and the calibrated navigation parameters to construct a unified environment perception model and provide input for subsequent path planning.

[0011] Further, spatio-temporal alignment synchronizes image frames and navigation data through timestamp matching and motion compensation algorithms, and utilizes the kinematic model of the unmanned ship to predict the trajectory of dynamic targets in the image. Feature-level fusion extracts edge, corner, and semantic information from the image, and combines speed and acceleration vectors in the navigation parameters to construct a comprehensive environment representation including static obstacles, dynamic target motion trends, and navigable areas.

[0012] Further, motion trend prediction and comprehensive collision risk assessment of dynamic targets include: Motion parameters of each dynamic target are extracted from dynamic target elements output by the environment perception model, including current position, speed vector, and acceleration vector; A motion trend prediction model is established to predict the motion process of dynamic targets, including short-term prediction for linear motion processes and long-term prediction for non-linear motion processes; Based on the predicted motion process of dynamic targets, in combination with the motion parameters of the unmanned ship, the dynamic relative distance between the target and the unmanned ship and the collision risk are obtained.

[0013] Further, the current position is obtained by projecting the pixel coordinates of the geometric center point of the target detection frame in the image to the geographic coordinate system through camera calibration, the speed vector is calculated based on the position change of the center point of the target detection frame in consecutive k frames of images combined with the time interval, and the acceleration vector is obtained by dividing the change in speed by the corresponding time interval difference when there are sufficient consecutive frames of images.

[0014] Further, when the target motion trajectory is approximately uniform linear in a short time, a short-term prediction method based on constant speed is adopted, and when the target motion presents nonlinear characteristics such as acceleration change or heading adjustment, a long short-term memory network is used to learn the nonlinear motion mode of the target, and the long-term dependence modeling capability of the long short-term memory network on time series data is used to capture the track evolution law of the dynamic target in a complex navigation environment.

[0015] Further, the threat evaluation indexes include: relative distance threat, reflecting the current and future minimum proximity of the target to the unmanned ship, the closer the distance, the higher the threat value, and the distance presents a negative exponential decay relationship; relative speed threat, indicating the approaching speed of the target to the unmanned ship, and the faster the approaching speed, the higher the collision risk; task conflict threat, reflecting the degree of motion trajectory of the target causing the interruption or large deviation of the original path of the unmanned ship task execution, and the task conflict threat is related to the deviation angle of the original path of the unmanned ship, the energy consumption and time loss required for path correction.

[0016] Further, the relative speed threat value is quantified based on the projection component of the relative speed of the two in the collision direction, the larger the projection value, the higher the threat, and the sigmoid function is used to map to the normalized interval.

[0017] Further, the deviation angle, energy consumption increment and time delay are normalized and weighted summed respectively, and then a threshold comparison method is used to judge whether a task conflict threat is generated.

[0018] Further, the optimization process of the multi-target conflict logic is: when the spatial avoidance areas of multiple high-threat targets overlap, the target with the highest threat evaluation index is preferentially selected for active avoidance, and secondary avoidance measures are taken for the second high-threat target; when the avoidance directions of all high-threat targets conflict, the minimum risk path strategy is started: the total threat exposure area corresponding to each possible avoidance direction is calculated, and the path with the smallest exposure area is selected as the optimal obstacle avoidance scheme.

[0019] The beneficial effects of the present application are: the unmanned ship path optimization method based on visual detection provided by the present application realizes real-time perception of the surrounding environment through visual detection, inputs parameters such as obstacle type, distance, relative speed and motion direction into a multi-dimensional threat evaluation function, realizes dynamic weighted sorting of different risk sources, and thus realizes navigation path optimization of the unmanned ship.

[0020] In addition to the purposes, features and advantages described above, the present application has other purposes, features and advantages. The present application will be further described in detail below with reference to the drawings. BRIEF DESCRIPTION OF DRAWINGS

[0021] The drawings constituting a part of the specification illustrate the exemplary embodiments of the present application and together with the description, serve to explain the present application. In the drawings: Figure 1 A schematic diagram of the overall flow of the unmanned ship path optimization method based on visual detection of the present application. DETAILED DESCRIPTION

[0022] It should be noted that the embodiments and features in the present application can be combined with each other without conflict. The present application will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0023] In order for those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings of the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should be within the scope of protection of the present application.

[0024] As shown in Figure 1 The present application provides an unmanned ship path optimization method based on visual detection, which is used to provide safe and efficient navigation path for unmanned ships in complex water environment, and can provide better navigation and obstacle avoidance strategy when multiple target obstacles are identified, so as to adjust the navigation trajectory in real time to avoid dynamic and static obstacles and ensure navigation safety. The method comprises the following steps: Step A: The data processing center receives the first image information collected by the visual sensor carried by the unmanned ship, and pre-processes the first image information, and receives the navigation parameter information from the unmanned ship itself, and calibrates the navigation parameter information, and constructs an environment perception model; In order to realize the path optimization of the unmanned ship, the surrounding environment information and the state data of the unmanned ship itself need to be obtained, wherein the environment information is the first image information, the first image information is collected by the visual sensor carried by the unmanned ship, the visual sensor is a sensor that can obtain the most original image to be processed by the vision processing system, and is a common intelligent sensor, which is widely used in navigation system, vision processing system. The first image information contains visual data of obstacles, channel boundaries and other dynamic targets in the water area. The pre-processing process includes image denoising, grayscale, contrast enhancement and other steps to improve image quality and reduce the computational burden of subsequent processing. In the path optimization of the unmanned ship, in addition to the need for collection and analysis of the external environment, the sailing state of the unmanned ship itself needs to be considered, so the sailing parameter information of the unmanned ship itself such as sailing speed, sailing direction, position coordinates and attitude angle is synchronously received to ensure that the path planning can be comprehensively judged in combination with real-time dynamic characteristics, wherein the sailing parameter information of the unmanned ship itself is provided by inertial measurement unit, GPS module and compass sensors in real time and calibrated through a data fusion algorithm, the data fusion algorithm includes a combination of Kalman filtering and complementary filtering to improve the accuracy and stability of the sailing parameters and effectively reduce the measurement error caused by a single sensor in a dynamic environment; Kalman filtering is a recursive algorithm for optimal estimation of state by using linear system state equation and statistical characteristics of system noise and observation noise, which can effectively fuse multi-source sensor data and suppress noise interference, and is especially suitable for the problems of attitude jitter and positioning drift of the unmanned ship in complex waters caused by wave disturbance; the complementary filtering further improves the real-time and accuracy of the attitude angle estimation by weighted fusion of high-frequency dynamic response and low-frequency steady-state accuracy through frequency domain characteristics; On this basis, the data processing center performs spatio-temporal alignment and feature-level fusion on the preprocessed image information and calibrated sailing parameters to construct a unified environment perception model and provide input for subsequent path planning, supporting the unmanned ship to realize high-precision environment understanding and autonomous decision-making in dynamic waters, wherein the spatio-temporal alignment refers to synchronizing and unifying the coordinates of the data collected by different sensors in time and space dimensions to ensure that the obstacle position in the image information and the current sailing position and direction of the unmanned ship are consistent, eliminating the perception deviation caused by sampling delay or coordinate system difference, specifically, the image frames and sailing data are synchronized through timestamp matching and motion compensation algorithm, and the kinematic model of the unmanned ship is used to predict the trajectory of the dynamic target in the image, thereby realizing dynamic perception; The feature-level fusion extracts the edges, corner points and semantic information in the image, combines the speed and acceleration vectors in the navigation parameters, constructs a comprehensive environment representation including static obstacles, dynamic target motion trends and navigable areas, and then provides a basis for subsequent path planning. For example, through feature-level fusion, the contour edges (such as HOG features) and semantic categories (such as "ship" and "floating object" labels detected by YOLOv8) of obstacles in the image can be extracted, and the current speed vector and heading angle of the unmanned ship in the navigation parameters can be combined to calculate the relative position and relative motion direction of the obstacle relative to the unmanned ship. The relative position is mapped from the image pixel coordinate system to the geographic coordinate system through coordinate transformation, and the relative motion direction is calculated by the time sequence displacement difference of the center point of the target detection box. The time sequence displacement difference refers to the change in pixel coordinates of the center point of the target detection box in consecutive two or more images, combined with the camera calibration parameters and the unmanned ship pose information, to calculate the actual moving direction and speed component in the geographic coordinate system, so as to judge the motion trend of the dynamic target. The information output by the environment perception model includes but is not limited to: static environmental elements such as channel boundaries, fixed obstacles and their positions and sizes; dynamic target elements such as real-time positions, speed vectors and predicted movement trends of moving obstacles; and navigable areas, which are the water area range that the unmanned ship can safely pass based on the calculation of obstacle distribution and water depth data. The calculation method can be to divide the remaining water space into regular grids after excluding the projection area of static obstacles and the safety distance buffer of dynamic targets, assign the passage cost based on the water depth layer and the dynamic obstacle avoidance radius, and generate a two-dimensional gridded navigable area map, thereby providing data for subsequent path planning. It should be noted that there are significant differences between water surface navigation and land navigation paths, mainly in that water surface navigation is more dynamic, the reference is unstable, such as the significant impact of water flow speed and wind direction changes on the track, and there is a lack of fixed road network constraints, so the existing path planning method based on roads cannot be directly applied to water surface environment. Step B: Based on the data output by the aforementioned environment perception model, the motion trend of the dynamic target is predicted and the collision risk of the unmanned ship is comprehensively evaluated. After building the environment perception model, various data directly collected can be output as structured environment information including static obstacle positions, dynamic target real-time poses and navigable areas, thereby supporting trajectory prediction and collision risk quantification analysis of moving targets. Specifically, the above process includes: Step B1: Extract the motion parameters of each dynamic target from the dynamic target elements output by the environment perception model, including the current position, speed vector and acceleration vector. The motion parameters of the dynamic target are important input basis for path planning of the unmanned ship, and are used to construct a perception model of surrounding traffic trends. The motion parameters of the dynamic target include a current position, a velocity vector, and an acceleration vector. The current position is obtained by projecting a geometric center point of a target detection frame in an image to a geographical coordinate system through camera calibration. The velocity vector is calculated based on a position change of the center point of the target detection frame in k consecutive image frames combined with a time interval. The acceleration vector is obtained by dividing a speed change amount by a corresponding time interval difference when sufficient consecutive image frames are available. In this embodiment, sufficient consecutive image frames refer to k being greater than or equal to 10 frames. In addition to dividing the speed change amount by the corresponding time interval difference, the acceleration estimation value can also be obtained by performing smoothing processing on the speed sequence based on polynomial fitting or Kalman filtering and then taking the derivative. In this embodiment, no limitation is made as long as the acceleration vector of the dynamic target can be obtained. Step B2: establishing a motion trend prediction model to predict the motion process of the dynamic target. The prediction process includes short-term prediction for linear motion process and long-term prediction for nonlinear motion process. Since the motion of the dynamic target in the water surface environment is often affected by wind, waves, water flow, and control instructions, it presents significant nonlinearity and uncertainty. Such nonlinear motion is difficult to accurately obtain through a traditional linear model, and it also includes conventional linear motion targets. Therefore, a motion trend prediction model needs to be established. The motion trend prediction model realizes prediction of the motion trend of linear dynamic targets through short-term prediction and prediction of the motion trend of nonlinear dynamic targets through long-term prediction. When the target moves in an approximately constant-speed straight line state in a short time, a short-term prediction method based on constant speed is adopted, for example, a Kalman filter is used to recursively estimate the position and speed, and a state vector, a state transition equation, and an observation update are used to realize smooth prediction of the trajectory. When the target motion presents nonlinear characteristics such as acceleration change or heading adjustment, a long short-term memory network is used to learn the nonlinear motion pattern of the target. The long short-term memory network has the ability to model long-term dependencies of time series data, and can capture the trajectory evolution law of the dynamic target in a complex navigation environment. For example, when the position and speed time series data of the target in a fixed time are input into the model, the model can automatically learn the motion pattern and output the position prediction sequence at each future time. The position prediction sequence at each future time is connected to form a complete prediction result of the future motion trajectory of the dynamic target. Step B3: based on the predicted motion process of the dynamic target and in combination with the motion parameters of the unmanned ship, obtaining a dynamic relative distance between the target and the unmanned ship and a collision risk. The predicted dynamic target trajectory is spatio-temporally aligned with the current heading, speed and position information of the unmanned ship, the minimum relative distance and the closest approach distance and time of the two in the subsequent time sequence are calculated, the size of the unmanned ship and the safety threshold are combined to determine whether there is a collision risk; if the predicted closest approach distance is less than the preset safety distance, it is determined that there is a collision threat, the obstacle avoidance decision mechanism is triggered, and the path planning adjustment stage is entered, wherein the closest approach distance (DCPA) and the approach time (TCPA) are calculated by the relative motion vector, which can be referred to in the prior art, and will not be expanded in this embodiment; Step C: constructing a multi-dimensional threat assessment function, quantifying and sorting all known collision risk sources, defining the threat assessment index of each risk source, and optimizing the multi-objective conflict logic; When there are multiple target crossing encounters, the collision risk level of the collision risk source needs to be quantified and sorted, so as to optimize the subsequent path planning process. In this embodiment, a multi-dimensional threat assessment function is constructed, and the threat assessment index of each risk source is defined. Specifically, the threat assessment index includes: Relative distance threat, reflecting the current and future minimum proximity of the target and the unmanned ship, the closer the distance, the higher the threat value, showing a negative exponential decay relationship. It can be calculated by combining the minimum relative distance in step B3 and the safety distance threshold, and for example, when the minimum relative distance of the known target in the future prediction period is d_min and the safety distance threshold is d_safe, the relative distance threat value can be defined as 1-d_min / d_safe, and when d_min≤d_safe, the threat value tends to 1; Relative speed threat, indicating the approach rate of the target and the unmanned ship, and the faster the approach speed, the higher the collision risk, and the relative speed threat value can be quantified based on the projection component of the relative speed of the two in the collision direction, the larger the projection value, the higher the threat, which can be mapped to the normalized interval using a sigmoid function; Task conflict threat, reflecting the degree to which the target motion trajectory causes the unmanned ship to interrupt the task execution or deviate significantly from the original path. The task conflict threat is related to the deviation angle of the original route of the unmanned ship, the energy consumption and time loss required for path correction. The deviation angle, energy consumption increment and time delay can be normalized and weighted summed respectively, and then the threshold comparison method is used to determine whether there is a task conflict threat; Based on the above three threat indicators, multi-dimensional threat analysis is performed on the obstacles, and the dimensions considered include but are not limited to: collision risk threat based on phase distance, dynamic threat based on relative speed, and task conflict threat, wherein the core comprehensive threat score for path optimization decision is mainly calculated through a cooperative evaluation model which integrates the two key parameters of relative distance threat and relative speed threat; for the task conflict threat, it is used as an independent judgment condition to identify high-priority conflict scenarios that may cause task interruption, so as to ensure that the unmanned ship can accurately identify the high-threat target that needs to be avoided most at present, and the specific formula is: T_S=(α / TTC +β / d )×(1+W×T), wherein T_S is the comprehensive threat score, TTC is the predicted time to collision, which is calculated from the relative distance d and the projection component of the relative speed in the collision direction, the smaller the TTC, the higher the collision urgency, d is the real-time relative distance between the unmanned ship and the risk source, W is the weighting coefficient of the task conflict threat, T is the task conflict threat value, wherein the task conflict threat value is only 0 or 1, T is 1 when there is significant task deviation or execution interruption, otherwise it is 0, α and β are normalized weight coefficients, satisfying α+β=1, used to balance the contribution degree of collision urgency and spatial proximity, W is the weight adjustment factor of task conflict, which is usually set to a positive value to improve the obstacle avoidance response sensitivity in high-priority task scenarios; when T=1, the comprehensive threat score is significantly amplified, thereby triggering the obstacle avoidance mechanism preferentially, when T=0, the comprehensive threat score is determined only by the collision urgency and the spatial proximity, ensuring that the obstacle avoidance priority can still be accurately evaluated without task interference, while avoiding over-response to low-risk targets; Wherein, the calculation method of the task conflict threat value T is: T=α·Δθ / θ_max+β·ΔE / E_base+γ·Δt / t_plan, wherein Δθ is the path deviation angle, θ_max is the maximum allowed deviation angle, ΔE is the energy consumption increment caused by path correction, E_base is the baseline energy consumption, Δt is the time delay, t_plan is the planned sailing time, α, β, γ are normalized weight coefficients, satisfying α + β + γ = 1, in the formula, Δθ / θ_max refers to the relative deviation degree, the closer this value is to 1, the more serious the heading deviation, and the greater the influence on the original task; ΔE / E_base is the relative energy consumption ratio, which processes the energy consumption increment caused by avoidance and the baseline energy consumption by ratio, reflecting the relative increase degree of energy consumption, the larger the ratio, the heavier the burden on the energy system caused by avoidance action; Δt / t_plan is the relative time delay rate, reflecting the time efficiency loss of task execution, the larger the value, the more significant the sailing delay, after weighted summation, the task conflict threat value T is obtained, when T≥0.5, it is determined that there is a significant task conflict threat, T takes the value of 1, otherwise it is 0; Among them, the normalization process is realized by scaling each index to the interval [0, 1], and before performing the normalization operation, the original data also needs to be input optimized to meet the requirements of the normalization process, including obtaining the statistical distribution characteristics of the original data such as maximum, mean, variance, etc., and filtering or smoothing the abnormal values according to the above, and then performing the normalization operation. In the normalization process of relative speed threat, the minimum-maximum normalization method is used to map the relative speed projection component to the interval [0, 1], and the sigmoid function is used for nonlinear enhancement to highlight the threat sensitivity at high closing speed. In the normalization process of task conflict threat, Z-score standardization combined with threshold truncation strategy is adopted to eliminate the dimension difference and suppress extreme value disturbance, so as to ensure the comparability of each threat component in fusion; Among them, the optimization process of multi-target conflict logic is: when the spatial avoidance areas of multiple high-threat targets overlap (for example, two ships are located in the left front and right front of the unmanned ship respectively, and both need to avoid to the middle), the target with the highest threat evaluation index is preferentially selected for active avoidance, and the secondary avoidance measure is taken for the secondary high-threat target. When the avoidance directions of all high-threat targets conflict, the minimum risk path strategy is started: calculate the total threat exposure area corresponding to each possible avoidance direction, and select the path with the smallest exposure area as the optimal obstacle avoidance scheme. Step D: Based on the output threat evaluation result, dynamically adjust the heading and speed of the unmanned ship, and generate a smooth obstacle avoidance trajectory combining environmental constraints and dynamics model.

[0025] According to the threat evaluation result of step C, the threat evaluation index of each risk source is updated in real time and input into the obstacle avoidance decision center of the unmanned ship. The obstacle avoidance decision center triggers the corresponding response strategy according to the threat level: for high-threat targets, an emergency avoidance path is planned immediately, the heading is adjusted to increase the relative distance, and the speed is controlled to shorten the exposure time; for low-threat targets, gradual heading adjustment is adopted to ensure smooth path and meet the dynamics constraints, and combined with environmental information such as wind and wave disturbance, channel boundary and static obstacles, the unmanned ship navigation path optimization in dynamic environment is realized. Among them, in order to make the obstacle avoidance trajectory smooth and prevent the unmanned ship from producing violent shaking due to frequent turning or sudden speed change, it is also necessary to generate a smooth obstacle avoidance trajectory combining environmental constraints and dynamics model, and to smooth the heading angle by using spline interpolation or Bezier curve, and to limit the acceleration change rate through speed profile planning to ensure smooth motion. For details, refer to the prior art, which will not be described in detail in this embodiment.

[0026] The above descriptions are only the preferred embodiments of the present application, and are not intended to limit the present application. The present application can have various modifications and changes for those skilled in the art. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A path optimization method for unmanned surface vessels based on visual detection, characterized in that: include: The data processing center receives the first image information collected by the visual sensor on the unmanned vessel, preprocesses the first image information, and receives the navigation parameter information from the unmanned vessel itself, calibrates the navigation parameter information, and then constructs an environmental perception model. Based on the data output by the aforementioned environmental perception model, the motion trend of dynamic targets is predicted and the collision risk of unmanned vessels is comprehensively assessed. Predicting the motion trend of dynamic targets and comprehensively assessing the collision risk of unmanned vessels includes: Extract the motion parameters of each dynamic target from the dynamic target elements output by the environmental perception model, including the current position, velocity vector, and acceleration vector; Establish a motion trend prediction model to predict the motion process of dynamic targets. The prediction process includes short-term prediction and long-term prediction, where short-term prediction is for linear motion processes and long-term prediction is for nonlinear motion processes. Based on the predicted motion process of the dynamic target, combined with the motion parameters of the unmanned vessel itself, the dynamic relative distance between the target and the unmanned vessel and the collision risk are obtained. The current position is obtained by back-projecting the pixel coordinates of the geometric center point of the target detection box in the image to the geographic coordinate system through camera calibration. The velocity vector is calculated based on the position change of the center point of the target detection box in k consecutive frames of images, combined with the time interval. The acceleration vector is obtained by dividing the velocity change when there are enough consecutive frames by the corresponding time interval difference. Construct a multi-dimensional threat assessment function to quantify and rank all known collision risk sources, define threat assessment indicators for each risk source, and optimize multi-objective conflict logic; Based on the output threat assessment results, the course and speed of the unmanned vessel are dynamically adjusted, and a smooth obstacle avoidance trajectory is generated by combining environmental constraints and dynamic models.

2. The unmanned surface vessel path optimization method based on vision detection according to claim 1, characterized in that: The data processing center performs spatiotemporal alignment and feature-level fusion of the preprocessed image information and calibrated navigation parameters to construct a unified environmental perception model, providing input for subsequent path planning.

3. The unmanned surface vessel path optimization method based on vision detection according to claim 2, characterized in that: Spatiotemporal alignment synchronizes image frames and navigation data through timestamp matching and motion compensation algorithms, and uses the kinematic model of the unmanned vessel to predict the trajectory of dynamic targets in the image. Feature-level fusion extracts edges, corners and semantic information from the image, and combines them with velocity and acceleration vectors from the navigation parameters to construct a comprehensive environmental representation that includes static obstacles, dynamic target motion trends and navigable areas.

4. The unmanned surface vessel path optimization method based on vision detection according to claim 1, characterized in that: When the target's trajectory is approximately uniform and straight for a short period of time, a short-term prediction method based on constant velocity is used. When the target's motion exhibits nonlinear characteristics such as acceleration changes or heading adjustments, a nonlinear motion pattern of the target based on a long short-term memory network is used. This network's ability to model long-term dependencies on time series data is utilized to capture the trajectory evolution patterns of dynamic targets in complex navigation environments.

5. The unmanned surface vessel path optimization method based on vision detection according to claim 1, characterized in that: Threat assessment metrics include: relative distance threat, which reflects the current and future minimum proximity of the target to the unmanned vessel; the closer the distance, the higher the threat value, exhibiting a negative exponential decay relationship; relative speed threat, which represents the approach rate of the target to the unmanned vessel; and mission conflict threat, which reflects the degree to which the target's trajectory will cause the unmanned vessel's mission execution to be interrupted or significantly deviate from its original path. This mission conflict threat is related to the deviation angle of the unmanned vessel's original route, the energy consumption required for path correction, and the time loss.

6. The unmanned surface vessel path optimization method based on vision detection according to claim 5, characterized in that: The relative velocity threat value is quantified based on the projection components of the relative velocities of the two objects in the collision direction. The larger the projection value, the higher the threat. The sigmoid function is used to map the values ​​to a normalized range.

7. The unmanned surface vessel path optimization method based on vision detection according to claim 5, characterized in that: The deviation angle, energy consumption increment, and time delay are normalized and weighted summed respectively, and then a threshold comparison method is used to determine whether a task conflict threat has occurred.

8. The unmanned surface vessel path optimization method based on vision detection according to claim 5, characterized in that: The optimization process of multi-target conflict logic is as follows: when the spatial avoidance areas of multiple high-threat targets overlap, the target with the highest threat assessment index is selected for active avoidance, and secondary avoidance measures are taken for the second highest threat target; when the avoidance directions of all high-threat targets conflict, the minimum risk path strategy is activated: calculate the total threat exposure area corresponding to each possible avoidance direction, and select the path with the smallest exposure area as the optimal obstacle avoidance scheme.

Images