A Speed-Based Collision Avoidance Method for Fishing Vessels Based on Risk Map Correction for Measurement Uncertainty

Abstract

This invention discloses a speed obstacle collision avoidance method for fishing vessels based on a risk map corrected for measurement uncertainty, belonging to the field of autonomous collision avoidance technology. The method includes: first, modeling the measurement uncertainty of the target vessel's state; then, constructing an initial spatial collision risk map based on the predicted encounter relationship between the vessel and the target vessel; next, correcting the initial risk map using measurement uncertainty parameters to obtain a corrected spatial collision risk map, and extracting high-risk contours from it to obtain a high-risk envelope region; then, constructing a speed obstacle region based on the high-risk envelope region and calculating a safe speed region; finally, selecting the optimal collision avoidance action from the safe speed region to generate control commands. This invention, by correcting the risk map for measurement uncertainty, enables the high-risk contours to dynamically reflect the impact of perception errors and encounter situations, achieving close coupling between risk characterization and collision avoidance decision-making, significantly improving the collision avoidance safety and robustness of fishing vessels under low-cost perception conditions.

Description

Technical Field

[0001] This invention belongs to the field of autonomous collision avoidance technology for ships, specifically relating to a speed obstacle avoidance method for fishing vessels based on a risk map corrected for measurement uncertainty, which is applicable to the safe autonomous navigation of fishing vessels under conditions of limited perception capabilities. Background Technology

[0002] With the continuous development of the intelligence level of fishing vessels, safe and reliable collision avoidance decision-making systems have become an important support for ensuring their safe operation in complex waters. However, due to limitations in economic costs, shipboard conditions, and equipment configuration capabilities, fishing vessels typically cannot be equipped with expensive, high-precision, and highly redundant multi-source sensing systems. This results in significant measurement errors and uncertainties in key information such as the target vessel's position, speed, and course acquired during navigation, thus affecting the reliability of collision avoidance decision-making algorithms.

[0003] Among existing autonomous collision avoidance technologies for ships, the speed barrier method is widely used in ship collision avoidance decision-making due to its advantages such as clear geometric expression, high local solution efficiency, and suitability for real-time collision constraint analysis. This method typically constructs a speed barrier region within a speed space based on the relative position and relative speed relationship between the ship and the target ship, and selects feasible speed or heading commands outside this region to achieve navigational safety. To construct the speed barrier region, it is usually necessary to geometrically expand the target ship's safety region and the ship's hull, combining the target ship's hull extent and its safety domain, to form a conflict domain for collision determination, and based on this, determine whether candidate speeds will pose a collision risk in the future time domain.

[0004] Currently, various methods have been proposed for characterizing the safety domain of ships, including circular, elliptical, and asymmetric domains. These methods can reflect the required safe distance around the target ship to a certain extent and play an important role in speed barrier design. However, collision hazards do not depend solely on the ship's safety domain itself, but are also closely related to the relative encounter situation between the ship and the target ship. Even when using the same or similar safety domain parameters, the spatial distribution, extension direction, and coverage of high-risk areas differ significantly when the target ship is in different encounter relationships such as port-side crossing, starboard-side crossing, head-on encounter, or overtaking. While traditional safety domains can provide the basic safety boundary around the target ship, they are still insufficient in expressing the spatial variation of the danger zone under different relative encounter situations, making it difficult to accurately reflect the actual collision risk distribution between the ship and the target ship under specific encounter relationships.

[0005] To improve risk characterization capabilities, some existing studies have proposed collision risk mapping methods. These methods represent collision hazards as a continuous distribution function on the navigation plane, describing the degree of danger at different spatial locations. Such methods can effectively reflect the positional changes of high-risk areas under different encounter situations and help express the overall risk distribution characteristics under multi-target vessel conditions. However, existing collision risk mapping methods are mostly used for risk visualization, and they typically lack tight coupling with collision avoidance decision-making algorithms. Furthermore, existing methods do not adequately consider target vessel state measurement errors and uncertainties, failing to fully reflect the impact of perception biases on the expansion, shift, and shape changes of risk areas.

[0006] Therefore, how to construct a collision avoidance decision-making method for fishing vessels that can fully consider the measurement errors and uncertainties of the target vessel's state has become a key technical problem that urgently needs to be solved in the current research on autonomous navigation of fishing vessels. Summary of the Invention

[0007] The purpose of this invention is to address the problems of existing collision avoidance methods failing to adequately reflect the spatial distribution differences of high-risk areas under different relative encounter situations, and the insufficient collision avoidance safety caused by target vessel state measurement errors and uncertainties. This invention provides a speed obstacle collision avoidance method for fishing vessels based on a risk map corrected for measurement uncertainty. By using a risk map corrected for measurement uncertainty, this invention enables high-risk profiles to dynamically reflect the impact of perception errors and encounter situations, achieving close coupling between risk characterization and collision avoidance decision-making, and significantly improving the collision avoidance safety and robustness of fishing vessels under low-cost perception conditions.

[0008] To achieve the above objectives, the present invention adopts the following technical solution: a method for collision avoidance of fishing vessel speed obstacles based on a risk map corrected for measurement uncertainty, comprising the following steps:

[0009] Step 1: Obtain the state information of the current ship and the target ship, perform measurement uncertainty modeling on the state information of the target ship, and obtain the measurement uncertainty parameters of the target ship;

[0010] Step 2: Based on the current status and predicted encounter relationship between the ship and the target ship, construct an initial space collision risk map;

[0011] Step 3: Correct the initial space collision risk map using measurement uncertainty parameters to obtain the corrected space collision risk map;

[0012] Step 4: Extract high-risk contours from the modified space collision risk map to obtain the high-risk envelope region;

[0013] Step 5: Construct a speed barrier region based on the high-risk envelope region, and calculate the safe speed region based on the candidate speed set of the ship;

[0014] Step 6: Select the optimal collision avoidance maneuver from the safe speed zone, generate and execute the ship's control commands.

[0015] Furthermore, step 1 specifically includes:

[0016] Step 1.1: Construct the short-time window state sequence of the target ship;

[0017] Step 1.2: Based on the short-time window state sequence, establish a short-time motion prediction model for the target ship;

[0018] Step 1.3: Calculate the state prediction residual of the target ship based on the short-time motion prediction model;

[0019] Step 1.4: Estimate the state uncertainty based on the state prediction residual, obtain the covariance matrix containing the uncertainties of the target ship's position, heading and speed, and perform regularization on the covariance matrix as a measurement uncertainty parameter.

[0020] Furthermore, step 2 specifically includes:

[0021] Step 2.1: Construct a discrete grid for the local navigation plane with the ship's current position as the origin;

[0022] Step 2.2: Calculate the predicted time for the ship to reach each grid point in the discrete grid from its current position;

[0023] Step 2.3: Predict the virtual state of the ship when it arrives at each grid point;

[0024] Step 2.4: Predict the state of each target ship at each grid point at the corresponding predicted time;

[0025] Step 2.5: Calculate the one-sided hazard assessment value based on the virtual state of the ship and the predicted state of the target ship;

[0026] Step 2.6: Based on the unilateral hazard assessment value, construct the initial spatial collision risk map for each target ship.

[0027] Step 2.7: When there are multiple target ships, the initial spatial collision risk maps of each single target are fused to obtain the initial spatial collision risk map of multiple targets, which is used as the initial spatial collision risk map.

[0028] Furthermore, step 3 specifically includes:

[0029] Step 3.1: Extract the uncertainty variances of the target ship's position, heading, and speed from the measurement uncertainty parameters;

[0030] Step 3.2: Propagate the state uncertainty to the planar position uncertainty to obtain the planar position uncertainty matrix;

[0031] Step 3.3: Construct the kernel function for the impact of uncertainty based on the planar position uncertainty matrix;

[0032] Step 3.4: Calculate the uncertainty correction gain based on the planar position uncertainty matrix;

[0033] Step 3.5: Use the uncertainty-affected kernel function and uncertainty correction gain to correct the initial space collision risk map, and obtain the single-target corrected space collision risk map;

[0034] Step 3.6: When there are multiple target ships, fuse the single-target corrected spatial collision risk maps to obtain a multi-target corrected spatial collision risk map, which serves as the corrected spatial collision risk map.

[0035] Furthermore, step 4 specifically includes:

[0036] Step 4.1: Set risk thresholds;

[0037] Step 4.2: Extract the grid points with risk values ​​greater than or equal to the risk threshold from the corrected space collision risk map into a comprehensive high-risk area point set;

[0038] Step 4.3: Based on the comprehensive high-risk area point set, determine the boundary point set of the comprehensive high-risk area;

[0039] Step 4.4: Extract contour lines, smooth and close the boundary point set to generate a high-risk profile;

[0040] Step 4.5: Define the closed region enclosed by the high-risk contour as the high-risk envelope region.

[0041] Furthermore, in step 4.3, the boundary point set of the comprehensive high-risk area is determined using the four-neighbor method.

[0042] Furthermore, in step 4.4, the generated high-risk profile is smoothed using a moving average method.

[0043] Furthermore, step 5 specifically includes:

[0044] Step 5.1: Based on the ship's safety expansion area, geometrically expand the high-risk envelope area to obtain a comprehensive effective high-risk area;

[0045] Step 5.2: Determine the candidate speed set based on the ship's permissible candidate speed range and candidate heading angle range;

[0046] Step 5.3: Determine whether the future trajectory corresponding to each candidate speed in the candidate speed set intersects with the comprehensive effective high-risk area, and take the intersecting candidate speeds as dangerous speeds to construct speed obstacle areas;

[0047] Step 5.4: Remove speed obstacle regions from the candidate speed set to obtain the safe speed region.

[0048] Furthermore, step 6 specifically includes:

[0049] Step 6.1: Construct a set of safe candidate actions based on the safe speed range;

[0050] Step 6.2: Construct a motion evaluation function that simultaneously considers the cost of navigation target deviation and the cost of motion smoothness;

[0051] Step 6.3: Select the candidate action that minimizes the action evaluation function value from the set of safe candidate actions as the optimal collision avoidance action;

[0052] Step 6.4: Convert the optimal collision avoidance maneuver into heading and speed control commands, and output them for execution after limiting the amplitude.

[0053] Furthermore, in step 6, when a collision risk is detected in the reference direction of the ship toward the target point, the desired avoidance command is calculated only once when the risk first occurs, and in subsequent control cycles, the ship smoothly approaches the desired avoidance command under the constraints of the rate of change of speed and the rate of change of heading, until the risk is eliminated and the ship resumes sailing toward the target point.

[0054] Compared with the prior art, the present invention has the following beneficial effects:

[0055] 1. Enhanced robustness to measurement errors: This invention models the measurement uncertainty of the target ship's position, heading, and speed information, and propagates the uncertainty to the space collision risk map. This enables the collision risk characterization results to more realistically reflect the potential high-risk occupation range of the target ship under perception error conditions, effectively improving the adaptability of collision avoidance decisions to measurement errors of low-cost perception equipment.

[0056] 2. Improved accuracy of risk area characterization: The present invention constructs a collision risk index based on the predicted encounter state of the ship and the target ship, and uses the influence of measurement uncertainty on the expansion, offset and shape change of the risk area to make it correct. This allows the high-risk profile to comprehensively reflect the combined effect of the risk superposition of multiple target ships, the relative encounter situation and measurement uncertainty, and overcomes the shortcomings of traditional fixed safety fields that cannot reflect the differences in encounter situation.

[0057] 3. Achieves close coupling between risk characterization and collision avoidance decision-making: This invention directly uses the high-risk contour of the modified space collision risk map to construct the speed obstacle region, so that the speed obstacle constraint no longer depends on the traditional fixed safety field or the current instantaneous state of the target ship, but dynamically reflects the actual high-risk boundary, thereby achieving effective unification of space risk perception and local real-time collision avoidance decision-making.

[0058] 4. Ensures the smoothness and feasibility of collision avoidance actions: Within the safe speed range, this invention comprehensively considers target maintenance and control smoothness, selects the optimal collision avoidance action, and adopts a control strategy of smooth command tracking and resumption of navigation after risk clearance. This effectively avoids frequent jumps in collision avoidance actions and improves the feasibility of the collision avoidance system under actual ship handling conditions.

[0059] Therefore, the method of the present invention has good engineering application prospects and can be widely applied to scenarios such as fishing vessel assisted navigation, intelligent collision avoidance and autonomous navigation. Attached Figure Description

[0060] Figure 1 A block diagram illustrating the implementation principle of the fishing vessel speed obstacle avoidance method based on a risk map corrected for measurement uncertainty, provided in an embodiment of the present invention.

[0061] Figure 2 This is a flowchart illustrating the calculation of the safe speed region in an embodiment of the present invention;

[0062] Figure 3 This is the initial spatial collision risk diagram of the first target ship in this embodiment of the invention;

[0063] Figure 4 This is the initial spatial collision risk diagram of the single target corresponding to the second target ship in this embodiment of the invention;

[0064] Figure 5 This is the integrated initial spatial collision risk map after multi-target fusion in this embodiment of the invention;

[0065] Figure 6 This is a corrected comprehensive space collision risk map considering the measurement uncertainty of the target ship in this embodiment of the invention;

[0066] Figure 7 This invention corrects the high-risk contour extraction results in the comprehensive space collision risk map in this embodiment;

[0067] Figure 8 This is a collision avoidance trajectory diagram of the ship in a dual-target ship encounter scenario according to an embodiment of the present invention. Detailed Implementation

[0068] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0069] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0070] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0071] This invention provides a speed obstacle collision avoidance method for fishing vessels based on a risk map corrected for measurement uncertainty. Its core is to construct a collision risk map that reflects the actual risk range of the target vessel, building upon the traditional speed obstacle method and incorporating the target vessel's state measurement error. Furthermore, it utilizes high-risk contours within the risk map to construct speed obstacle regions that better reflect the actual risk distribution. Simultaneously, it combines the speed obstacle method with local solutions for candidate speed or heading commands, enabling dynamic adjustment of the fishing vessel's collision avoidance strategy under measurement uncertainty conditions. This achieves high-safety and high-robustness collision avoidance decision-making for fishing vessels under low-cost perception conditions and complex encounter scenarios. Figure 1 As shown, the method of the present invention mainly includes the following steps:

[0072] Step 1: Obtain the state information of the current ship and the target ship, perform measurement uncertainty modeling on the state information of the target ship, and obtain the measurement uncertainty parameters of the target ship;

[0073] Step 2: Based on the current status and predicted encounter relationship between the ship and the target ship, construct an initial space collision risk map;

[0074] Step 3: Correct the initial space collision risk map using measurement uncertainty parameters to obtain the corrected space collision risk map;

[0075] Step 4: Extract high-risk contours from the modified space collision risk map to obtain the high-risk envelope region;

[0076] Step 5: Construct a speed barrier region based on the high-risk envelope region, and calculate the safe speed region based on the candidate speed set of the ship;

[0077] Step 6: Select the optimal collision avoidance maneuver from the safe speed zone, generate and execute the ship's control commands.

[0078] For step 1, after obtaining the target ship's current state information and the state information of the target ship at several recent moments, the system models the uncertainties of the target ship's position, heading, longitudinal speed, lateral speed, and bow roll rate based on the predicted residuals of the target ship's state within a short time window. This provides state inputs and uncertainty parameters for the subsequent construction of the collision risk map. The target ship state vector is defined as follows:

[0079] (1)

[0080] in, Indicates the first The target ship at the moment The state vector; and They represent the first The target ship's position coordinates; Indicates the first The target ship's heading angle; , and They represent the first The longitudinal speed, lateral speed, and bow roll rate of the target ship.

[0081] Set length as The short time window, then the first The target ship at the current moment Corresponding historical state sequence It can be represented as:

[0082] (2)

[0083] The short time window is used to characterize the recent state changes of the target ship. The specific modeling process is as follows.

[0084] Step 1.1: Construct the short-time window state sequence of the target ship

[0085] For the Target ship, extract the latest The state information at each moment forms a short-time window state sequence. To avoid crossing the heading angle... and An angle jump occurs, affecting the heading angle. The process involves expanding the data to ensure continuous change within a short time window.

[0086] Step 1.2: Establish a short-time motion prediction model for the target ship.

[0087] Within a short time window, assuming the first If the target ships move at approximately uniform speed and uniform angular velocity between adjacent sampling times, then according to the time... State versus time The state is then predicted in one step to obtain the predicted state vector. :

[0088] (3)

[0089] in, , , , , , They represent the first The target ship at the moment The predicted northward coordinates, eastward coordinates, heading angle, longitudinal velocity, lateral velocity, and bow roll rate are calculated as follows:

[0090] (4)

[0091] in, This indicates the time interval between adjacent sampling times.

[0092] Step 1.3: Calculate the target ship state prediction residuals

[0093] Based on time Actual state observations With time Predicted state values Calculate the state prediction residual vector :

[0094] (5)

[0095] Step 1.4: Estimate state uncertainty based on predicted residuals

[0096] Within a short time window, for the first The predicted residuals of the target ship at each time step are statistically analyzed to construct the target ship state uncertainty matrix. First, the mean of the predicted residuals within the window is calculated. :

[0097] (6)

[0098] Then calculate the sample covariance matrix of the predicted residuals. :

[0099] (7)

[0100] To avoid singularity or degradation of the covariance matrix when the residuals are small or the sample length is short, a regularization term is added to the covariance matrix to obtain a corrected uncertainty matrix. :

[0101] (8)

[0102] in, It is the identity matrix. This is the preset regularization coefficient.

[0103] Wherein, the state vector The uncertainty matrix is ​​used to characterize the current motion state of the target ship. It is used to characterize the measurement uncertainty of the target ship in terms of current position, heading, longitudinal speed, lateral speed and bow roll rate, and serves as the input basis for subsequent collision risk map construction and correction.

[0104] For step 2, based on the current state information of the ship and the target ship, as well as the modeling results of the measurement uncertainty of the target ship, the correction of the risk map due to measurement uncertainty is not considered at the moment. Only based on the current deterministic state between the ship and the target ship and the predicted encounter relationship, an initial spatial collision risk map is constructed, which serves as the basic input for the measurement uncertainty correction in the subsequent step 3. The specific process is as follows.

[0105] Step 2.1: Construct a local navigation plane discrete mesh

[0106] Establish a local north-east coordinate system with the current position of the ship as the origin, and construct a discrete grid within a given range. Each grid point is represented as .

[0107] in:

[0108] (9)

[0109] In the formula, These represent the ship's current north and east coordinates, respectively. These represent the offsets of the grid points relative to the ship's current position in the north and east directions, respectively. Representing grid points respectively coordinate.

[0110] Step 2.2: Calculate the predicted arrival time of the ship at the grid points.

[0111] Let the current speed modulus of this ship be... Then the ship will reach the grid point from its current position. The distance is :

[0112] (10)

[0113] Corresponding arrival time for:

[0114] (11)

[0115] Step 2.3: Predict the state of the target ship at the corresponding time.

[0116] This vessel has reached the grid point. Predicted heading at time Defined as:

[0117] (12)

[0118] Based on this, construct the virtual state of the ship at the grid points. :

[0119] (13)

[0120] Step 2.4: Predict the state of the target ship at the corresponding time.

[0121] Let the first The status of each target ship at the corresponding time for:

[0122] (14)

[0123] The velocity components of the target ship in the global coordinate system are:

[0124] (15)

[0125] Assuming the target ship is in time If the speed and heading remain constant, then its predicted position is:

[0126] (16)

[0127] Thus, the predicted state of the target ship is obtained:

[0128] (17)

[0129] In the formula, Indicates the first Current status of the target vessel; Indicates the target ship's heading; These represent the target ship's longitudinal velocity, lateral velocity, and bow roll rate, respectively. These represent the velocity components of the target ship in the north and east directions, respectively; This indicates the state of the target ship at the corresponding predicted time.

[0130] Step 2.5: Calculate the one-sided hazard assessment value

[0131] First, calculate the position of the target ship relative to your own ship and project it onto your own ship's heading coordinate system:

[0132] (18)

[0133] Similarly, calculate the position of this ship relative to the target ship and project it onto the target ship's heading coordinate system:

[0134] (19)

[0135] In the formula, These represent the lateral and longitudinal positions of the target vessel relative to the vessel itself, respectively. These represent the lateral and longitudinal positions of the vessel relative to the target vessel, respectively; in the lateral position, the starboard side is positive, and in the longitudinal position, the forward direction is positive.

[0136] Furthermore, let the forward, aft, port, and starboard axis lengths in the asymmetric ship safety domain be as follows:

[0137] (20)

[0138] in, Indicates the reference captain; These represent the forward, aft, port, and starboard dimension coefficients, respectively. These represent the forward, aft, port, and starboard axis lengths in the asymmetric ship safety domain, respectively.

[0139] Define a one-sided hazard assessment function for:

[0140] (twenty one)

[0141] Therefore, we can conclude that:

[0142] (twenty two)

[0143] In the formula, This indicates the hazard assessment value of the target vessel relative to the vessel itself on one side. This indicates the hazard assessment value of this vessel relative to the target vessel on one side.

[0144] Step 2.6: Construct the initial spatial collision risk map for a single target

[0145] Based on the unilateral hazard assessment value in step 2.5 The target ship is at the grid point Initial collision risk value of a single target at the location Defined as:

[0146] (twenty three)

[0147] By traversing all grid points within the local navigation plane, the first... Initial spatial collision risk map for each target ship.

[0148] Step 2.7: Construct a multi-target initial spatial collision risk map

[0149] When multiple target ships exist around the vessel, the initial risk maps of each target ship are fused to obtain a comprehensive initial spatial collision risk map of multiple targets. The fusion method is defined as follows:

[0150] (twenty four)

[0151] in, This indicates the number of target vessels participating in the collision avoidance decision-making process at the current moment; Represents grid points The initial collision risk value is calculated by considering all target vessels.

[0152] For step 3, based on steps 1 and 2, the impact of state measurement errors such as the target ship's position, heading, and speed on the target ship's potential occupation area is mapped onto the local navigation plane. This allows the collision risk map to not only reflect the deterministic encounter risk between the ship and the target ship, but also the expansion, shift, and shape changes of high-risk areas caused by perception errors and short-term state fluctuations. The specific process is as follows.

[0153] Step 3.1: Extract the uncertainty parameters of the target ship's state

[0154] Based on the output of step 1, let the first... The uncertainty matrix of the current state of the target ship is as follows: The diagonal terms of the uncertainty matrix are denoted as:

[0155] (25)

[0156] in, , , , , , They represent the first Uncertainty variance of the north coordinates, east coordinates, heading angle, longitudinal speed, lateral speed and bow roll rate of the target ship; This represents the diagonal matrix construction operator.

[0157] Because in step 2, for any grid point The predicted time for this ship to arrive at that point has been calculated. Therefore, the uncertainty of the bow roll rate can be further converted into an additional impact on the uncertainty of the heading prediction, and grid points can be defined. The corresponding effective heading variance of the target ship for:

[0158] (26)

[0159] Therefore, construct the first The target ship is at the grid point The corresponding simplified state uncertainty matrix :

[0160] (27)

[0161] Step 3.2: Propagate the state uncertainty to the planar position uncertainty

[0162] To map the uncertainty of the target ship's state into a spatial risk distribution, it is necessary to first propagate the uncertainty in the state space to the planar position space. Based on the short-time motion prediction model of the target ship in step 2, the target ship is then positioned at grid points. Predicted position at corresponding time Represented as:

[0163] (28)

[0164] In the formula, Indicates the first The target ship was Time Prediction coordinate; Indicates the first The target ship was Time Prediction coordinate.

[0165] To propagate state uncertainty to position uncertainty, equation (16) is linearized to the first order, yielding the Jacobian matrix of position with respect to state. :

[0166] (29)

[0167] Therefore, the first The target ship is at the grid point Uncertainty matrix of planar position at corresponding time It can be represented as:

[0168] (30)

[0169] To avoid singularity or degeneracy issues in the numerical calculation of the position uncertainty matrix, a regularization term is added to it, resulting in a corrected position uncertainty matrix. :

[0170] (31)

[0171] in, These are preset positive numbers.

[0172] Step 3.3: Construct the kernel function for the impact of uncertainty

[0173] No. The target ship is at the grid point The spatial uncertainty at the corresponding prediction time affects the kernel Defined as:

[0174] (32)

[0175] Step 3.4: Calculate the uncertainty correction gain

[0176] To ensure that the degree of correction of the initial space collision risk map by measurement uncertainty matches the magnitude of the target ship's state uncertainty, an uncertainty correction gain is introduced. The target ship is at the grid point Uncertainty correction gain at corresponding time for:

[0177] (33)

[0178] in, This is the gain adjustment coefficient. This represents the matrix trace operation.

[0179] Step 3.5: Construct a single-target corrected spatial collision risk map

[0180] Based on the initial spatial collision risk map of the single target obtained in step 2 And by combining the influence kernel of spatial uncertainty and the uncertainty correction gain, the first... Corrected spatial collision risk value for each target ship. For grid points. Its corrected risk value is defined as:

[0181] (34)

[0182] in, Indicates the first The target ship is at the grid point The corrected space collision risk value.

[0183] Step 3.6: Construct a multi-objective corrected spatial collision risk map

[0184] When multiple target ships exist simultaneously around the vessel, the modified spatial collision risk maps corresponding to each target ship are fused to obtain a comprehensive modified spatial collision risk map for multiple targets. The fusion method is the same as in step 2.7.

[0185] For step 4, based on the spatial collision risk map obtained in step 3 after considering measurement uncertainties, a high-risk contour is extracted from the corrected spatial collision risk map. This contour characterizes the high-risk occupancy boundary of the target ship within the current moment and short-term prediction range, and provides geometric boundary input for the speed barrier region construction in subsequent step 5. The high-risk contour is not a traditional fixed safety domain boundary, but rather a set of spatial points in the collision risk map that reach a given risk threshold. It comprehensively reflects the range of the high-risk area under the combined influence of the relative encounter situation and measurement uncertainties. The specific process is as follows:

[0186] Step 4.1: Determine the high-risk threshold

[0187] To identify high-risk areas from the comprehensive space collision risk map, a risk threshold is first set. This is used to distinguish between high-risk and non-high-risk areas. The risk threshold satisfies... When the comprehensive space collision risk value is greater than or equal to the risk threshold. When the overall spatial collision risk value is less than the aforementioned risk threshold, the corresponding spatial location is identified as a high-risk area. At that time, the corresponding spatial location was determined to be a non-high-risk area.

[0188] Step 4.2: Extract the comprehensive high-risk area point set

[0189] Based on the multi-objective corrected space collision risk map obtained in step 3 The overall high-risk area at the current moment is defined as:

[0190] (35)

[0191] in, This indicates that, at the current moment, the overall spatial collision risk value among all grid points is not less than the preset risk threshold. The set of points, i.e., the comprehensive high-risk area within the current local navigation plane.

[0192] Step 4.3: Determine the set of boundary points for the comprehensive high-risk area

[0193] Due to the comprehensive high-risk areas It is still a discrete set of grid points and cannot be directly used as a geometric boundary for subsequent velocity barrier region construction. Therefore, it is necessary to further extract its boundary point set.

[0194] Let the set of boundary points of the comprehensive high-risk area be . .in, Represents grid points The set of neighborhood points; Represents grid points It does not belong to the overall high-risk area. In one implementation method, the boundary can be determined using a four-neighborhood approach.

[0195] Step 4.4: Generate high-risk profiles

[0196] To transform discrete boundary point sets into continuous geometric boundaries, high-risk contours are obtained directly using contour line extraction from a comprehensive spatial collision risk map. ,Right now:

[0197] (36)

[0198] in, This represents the spatial collision risk distribution function that is a comprehensive multi-objective correction at the current moment.

[0199] Step 4.5: Contour Smoothing and Closure Processing

[0200] Since the comprehensive space collision risk map is calculated from discrete grid points, the extracted high-risk contours may exhibit local jaggedness, boundary discontinuities, or local burrs. To improve the stability of subsequent velocity barrier region construction, the high-risk contours are smoothed and closed.

[0201] Let the original high-risk profile curve be represented as .in, The first high-risk profile Discrete boundary points, Indicates the number of boundary points.

[0202] The contour point sequence is smoothed by filtering to obtain a smoothed high-risk contour. .in, The contour smoothing operator can be implemented using the moving average method.

[0203] If the contour point set fails to form a closed boundary due to local holes or discretization errors, further closure processing is performed to obtain the final high-risk contour:

[0204] (37)

[0205] in, This represents the contour closure operator.

[0206] After smoothing and closing processing This is the comprehensive high-risk profile that will ultimately be used for the construction of the speed barrier area at the current moment.

[0207] Step 4.6: Construct the risk envelope region corresponding to the high-risk profile.

[0208] To facilitate geometric expansion and velocity barrier region construction in subsequent step 5, the final high-risk profile can be... The enclosed region is defined as the high-risk envelope region at the current moment. :

[0209] (38)

[0210] in, It represents the internal area enclosed by the outline.

[0211] For step 5, based on step 4, a speed barrier region is constructed to describe the set of candidate speeds that would cause the vessel to enter a high-risk area within a given prediction time domain, and a safe speed region is calculated. The calculation process is as follows: Figure 2 As shown. The specific process is as follows.

[0212] Step 5.1: Construct the ship's safety expansion zone

[0213] Due to the comprehensive high-risk envelope region obtained in step 4 This mainly describes the high-risk area range of the ship's reference point in the local navigation plane. In order to further consider the ship's hull length in the speed obstacle area construction, it is necessary to safely expand the high-risk envelope area.

[0214] Establish the ship's safe expansion zone Defined as:

[0215] (39)

[0216] in, This represents the equivalent safety radius formed by the ship's reference point, the outer edge of the hull, and additional safety margins.

[0217] Based on the aforementioned safety expansion area of ​​the vessel, the comprehensive high-risk envelope area obtained in step 4 is... Geometric expansion yields a comprehensive and effective high-risk area:

[0218] (40)

[0219] in, Represents the Minkowski sum operation. The boundary contour of the overall effective high-risk region is... .in, Represents the region boundary operator.

[0220] Step 5.2: Determine the candidate velocity space

[0221] Let the range of candidate speeds allowed for this ship at the current decision moment be . The allowed range of candidate heading angle values ​​is: Then any candidate control action can be represented as:

[0222] (41)

[0223] in, Indicates the candidate speed. Indicates the candidate heading angle.

[0224] The candidate velocity vector is represented as .

[0225] Therefore, the set of candidate velocities for this ship is: .

[0226] Step 5.3: Calculate the speed obstacle zone

[0227] The comprehensive high-risk envelope area is obtained through step 4. Based on this, for any candidate velocity vector If the ship follows the candidate speed The future trajectory of movement enters a comprehensive and effective high-risk area. If a candidate speed is deemed too high, it is considered to lead the vessel into a high-risk zone and is thus classified as a dangerous speed. Therefore, the speed barrier zone is defined as follows:

[0228] (42)

[0229] in This represents the speed obstacle zone at the current moment, which is the set of all candidate speeds that would cause the ship's future trajectory to intersect with the overall effective high-risk zone.

[0230] Step 5.4: Calculate the safe speed zone

[0231] After obtaining the overall speed obstacle zone, all dangerous speeds are further eliminated from the candidate speed set of the vessel to construct a safe speed zone. The safe speed zone is defined as follows:

[0232] (43)

[0233] in, This represents the set of candidate speeds that will not cause the ship to enter the overall high-risk zone at the current moment.

[0234] For step 6, based on the safe speed range obtained in step 5, the optimal candidate speed is selected from the safe speed range and further converted into heading and speed control commands that the ship can execute, thereby realizing the output and execution of collision avoidance actions. The specific process is as follows.

[0235] Step 6.1: Construct a set of security candidate actions

[0236] Based on the safe speed zone obtained in step 5 Candidate collision avoidance actions are represented as The corresponding velocity vector is The set of safe candidate actions is defined as follows: .

[0237] Step 6.2: Construct the action evaluation function

[0238] To select the optimal action from the set of safety candidate actions, an evaluation function for the candidate action is constructed. This evaluation function considers both the deviation between the candidate action and the desired navigation state, as well as the smoothness of the action change, and is defined as follows:

[0239] (44)

[0240] in, This indicates the cost of deviation from the navigation target. Indicates the cost of smoothness of motion; and These represent the weighting coefficients of the two costs, respectively.

[0241] Let the reference velocity vector corresponding to the ship's current mission be... ,in These are the reference speed magnitude and reference heading, respectively. The target deviation cost is then defined as:

[0242] (45)

[0243] Suppose that the speed and heading commands executed on this ship at the last control moment are respectively and Then the cost of smoothness of motion is defined as:

[0244] (46)

[0245] in, and These represent the smoothness weights for the speed change term and the heading change term, respectively.

[0246] Step 6.3: Solve for the optimal collision avoidance action

[0247] In the set of security candidate actions Within this process, an evaluation function is calculated for all candidate actions. The candidate action with the smallest evaluation value is selected as the optimal collision avoidance action at the current moment. .

[0248] Let the optimal collision avoidance action be expressed as .in, This indicates the optimal collision avoidance speed at the current moment. This indicates the optimal collision avoidance heading at the current moment.

[0249] Step 6.4: Control command output and execution

[0250] The optimal collision avoidance maneuver is converted into a control command executable by the ship's control system, defined as:

[0251] (47)

[0252] To meet the ship's maneuverability constraints, the output control commands can be limited. Let the maximum permissible rate of change of course be... The maximum permissible rate of change of speed is The control command after limiting can be expressed as:

[0253] (48)

[0254] in, This represents the saturation limiting function. The control commands are sent to the ship's heading controller and speed controller for execution, thereby completing the collision avoidance action output for the current control cycle.

[0255] The present invention will now be described in further detail.

[0256] The following section uses a dual-target vessel encounter scenario to further illustrate the proposed collision avoidance method for fishing vessels based on a risk map corrected for measurement uncertainty. It should be noted that this embodiment is only used to illustrate the technical solution of the present invention and does not constitute a limitation on the scope of protection of the present invention.

[0257] In this embodiment, the ship's state vector includes its northward position, eastward position, heading angle, longitudinal velocity, lateral velocity, and bow roll rate. Initially, the ship is located at the origin of the local navigation coordinate system, with a heading angle of 0°, a longitudinal velocity of 5.0 m / s, and both the lateral velocity and bow roll rate are 0. The target point is set 150 m directly in front of the ship's initial position, i.e., the target point's northward coordinates are 150 m, and its eastward coordinates are 0 m.

[0258] This embodiment sets up two target ships, denoted as TS1 and TS2. The first target ship, TS1, is located in the port forward region of the main ship's OS, sailing eastward at a speed of approximately 5.0 m / s. Its four most recent historical positions are approximately 60 m north, 40 m east, 35 m east, 30 m east, and 25 m north, respectively. The second target ship, TS2, is located slightly to the right forward of the main ship's OS, sailing southward at a speed of approximately 4.0 m / s. Its four most recent historical positions are approximately 5 m east, 80 m north, 76 m north, 72 m north, and 68 m north, respectively.

[0259] Considering that low-cost sensing equipment on fishing vessels often suffers from measurement errors in practical applications, this embodiment incorporates measurement perturbations based on the target vessel's actual state to form a target vessel observation history sequence. Specifically, the standard deviations for position measurement error are set to 0.3 m, heading angle measurement error to 0.8°, longitudinal velocity measurement error to 0.08 m / s, lateral velocity measurement error to 0.03 m / s, and bow roll rate measurement error to 0.1° / s. The system calculates the one-step prediction residual of the target vessel based on the observation states at recent times and estimates the uncertainty in the target vessel's state measurement accordingly.

[0260] In the initial collision risk map construction phase, a local north-east navigation plane is established with the ship's current position as the origin. The local grid's northward range is set to [−100, 160] m, and its eastward range is set to [−100, 160] m, with a grid resolution of 2 m. For each grid point in the local navigation plane, the system first calculates the time required for the ship to reach that grid point and predicts the target ship's position at the corresponding time under the assumption of maintaining speed and heading. Subsequently, based on the ship's virtual state and the target ship's predicted state, single-target initial spatial collision risk maps are constructed for the first and second target ships, respectively, as follows: Figure 3 and Figure 4 As shown. By Figure 3 and Figure 4 It can be seen that, due to differences in initial position, direction of motion, and relative encounter situation, the spatial distribution of high-risk areas formed by different target ships in the local navigation plane also varies significantly.

[0261] Furthermore, the system fuses the initial spatial collision risk maps of the two target ships to obtain a comprehensive initial spatial collision risk map, such as... Figure 5 As shown. Figure 5 It reflects the overall collision risk distribution around the ship under the combined action of multiple target ships, providing a basis for subsequent measurement uncertainty correction and high-risk area extraction.

[0262] During the risk map correction phase, the system propagates the measurement uncertainty of the target ship's state to the predicted position plane of the target ship, and corrects the integrated initial space collision risk map based on the predicted position uncertainty, resulting in a corrected integrated space collision risk map that considers measurement uncertainties, such as... Figure 6 As shown. Compared to Figure 5 , Figure 6 It can further reflect the impact of target ship position, heading and speed measurement errors on the range and boundary shape of high-risk areas, making the risk area representation more in line with the actual collision avoidance requirements under low-cost perception conditions.

[0263] In the high-risk contour extraction stage, this embodiment sets the risk threshold to 0.35. When the risk value of a grid point in the corrected comprehensive risk map is not less than this threshold, it is identified as a high-risk point. All high-risk points form a comprehensive high-risk region, and the boundary contour of the high-risk region is further extracted. The resulting high-risk contour is as follows: Figure 7 As shown. This contour is used to describe the main high-risk boundaries in the revised risk map, providing geometric constraints for subsequent speed obstacle region construction. During the speed obstacle region construction stage, the system generates effective high-risk regions for speed obstacle judgment based on the high-risk regions and the ship's safety margin. In this embodiment, the ship's safety expansion radius is set to 6.0 m. The candidate speed range is set to 0~8 m / s, with a speed interval of 0.5 m / s; the candidate heading angle coverage... The heading angle interval is 5°. For each candidate speed and heading combination, the system determines whether the ship's future trajectory along that speed direction will enter the effective high-risk area; if it will, it is determined as a dangerous speed, otherwise it is determined as a safe speed.

[0264] During the collision avoidance maneuver calculation phase, the system first determines whether the reference speed pointing towards the target point is within the speed barrier area. If the target point direction is safe, the vessel continues to navigate in the target point direction; if there is a collision risk in the target point direction, the system selects the candidate speed with the smaller overall cost from the set of safe speeds as the desired avoidance command.

[0265] To avoid frequent changes in collision avoidance maneuvers, this embodiment employs an execution method of "maintaining the avoidance command - smooth tracking - resuming navigation after the risk is cleared." When a collision risk is detected in the direction of the target point, the system calculates the desired avoidance speed and heading only once when the risk first appears. Subsequently, the vessel no longer updates the avoidance command every moment, but instead smoothly approaches the desired avoidance command under the constraints of the rate of change of speed and the rate of change of heading. In this embodiment, the maximum rate of change of heading is set to 8 ∘ per control cycle, the maximum rate of change of speed is set to 0.3 m / s per control cycle, and the control cycle is set to 1 s.

[0266] During the avoidance process, the system continuously monitors whether the target point's direction has left the speed barrier area and determines whether the maximum risk value in the corrected integrated spatial collision risk map is lower than the preset risk clearance threshold. In this embodiment, the risk clearance threshold is set to 0.30, and the system requires that the safety conditions be met for several consecutive control cycles before entering the return-to-course state. When the return-to-course conditions are met, the ship recalculates the reference heading towards the target point based on its current position and gradually adjusts its speed and heading under the constraints of the rate of change of speed and the rate of change of heading to achieve a smooth return to course. If the target point's direction re-enters the speed barrier area during the return-to-course process, the system re-enters the avoidance and hold state.

[0267] The navigation trajectories of this vessel and the two target vessels during the aforementioned closed-loop collision avoidance process are as follows: Figure 8 As shown. By Figure 8 It can be seen that in the scenario of encountering two target ships, the ship can identify high-risk areas based on the revised integrated spatial collision risk map and generate collision avoidance commands by combining the speed obstacle method, thereby avoiding the high-risk areas formed by the target ship; when the risk is eliminated, the ship can gradually adjust its course and continue to sail towards the target point.

[0268] By incorporating the uncertainty of target vessel state measurement into the construction of spatial collision risk maps and the generation of speed obstacle regions, this invention significantly improves the collision avoidance safety of fishing vessels under low-cost perception conditions. This method addresses issues in existing collision avoidance studies based on speed obstacle methods, such as the inability of traditional safety approaches to fully reflect the spatial distribution differences of high-risk areas under different relative encounter situations and the failure of algorithms due to target vessel state measurement errors. By constructing an initial spatial collision risk map based on a collision risk index and further refining the risk map by incorporating the uncertainty of the target vessel's state, the collision risk characterization results more realistically reflect the potential high-risk area occupied by the target vessel, thereby enhancing the effectiveness of risk modeling and collision avoidance decision-making in complex encounter scenarios.

[0269] Specifically, this invention first utilizes the target ship state prediction residuals within a short time window to model the measurement uncertainty of the target ship's position, heading, and speed information, thereby avoiding misjudging the target ship's normal motion trend as measurement error. Based on this, a collision risk index is calculated based on the predicted encounter state of the ship and the target ship, constructing an initial spatial collision risk map. Furthermore, the uncertainty of the target ship's state is used to map the impact of measurement error and short-term fluctuations on the expansion, shift, and shape changes of the risk area onto the planar risk distribution. Subsequently, high-risk contours and high-risk envelope regions are extracted from the corrected comprehensive spatial collision risk map and mapped onto the velocity domain to construct a speed barrier region. This ensures that the speed barrier constraint no longer relies solely on the traditional fixed safety domain or the target ship's current instantaneous state, but simultaneously reflects the superposition of risks from multiple target ships, the relative encounter situation, and the impact of measurement uncertainty on the high-risk region boundary. Finally, within the safe speed region, considering both target maintenance and control smoothness, the optimal collision avoidance maneuver is selected, and heading and speed control commands are output, thereby improving the rationality and executability of the collision avoidance maneuver output.

[0270] In summary, the method of this invention improves the adaptability of risk characterization to perception errors by modeling measurement uncertainty, enhances the ability to characterize high-risk areas in complex multi-target ship encounter environments through spatial collision risk map correction and high-risk contour extraction, and achieves local real-time collision avoidance decision-making through coupling with the velocity barrier method. This results in higher collision avoidance safety for fishing vessels under the combined effects of measurement uncertainty and complex traffic environments. This method has promising engineering application prospects and can be widely applied to scenarios such as assisted navigation, intelligent collision avoidance, and autonomous navigation of fishing vessels.

[0271] This embodiment also provides a computer device, including: at least one processor, at least one memory, and computer program instructions stored in the memory, which implement the above-described method when executed by the processor.

[0272] This embodiment also provides a computer-readable storage medium storing computer program instructions that, when executed by a processor, implement the above-described method.

[0273] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0274] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0275] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1The function specified in one or more boxes.

[0276] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable apparatus for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0277] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for collision avoidance of a fishing vessel based on a risk map modified by measurement uncertainty, characterized in that, Includes the following steps: Step 1: Obtain the state information of the current ship and the target ship, perform measurement uncertainty modeling on the state information of the target ship, and obtain the measurement uncertainty parameters of the target ship; Step 2: Based on the current status and predicted encounter relationship between the ship and the target ship, construct an initial space collision risk map; Step 3: Correct the initial space collision risk map using measurement uncertainty parameters to obtain the corrected space collision risk map; Step 4: Extract high-risk contours from the modified space collision risk map to obtain the high-risk envelope region; Step 5: Construct a speed barrier region based on the high-risk envelope region, and calculate the safe speed region based on the candidate speed set of the ship; Step 6: Select the optimal collision avoidance maneuver from the safe speed zone, generate and execute the ship's control commands.

2. The fishing vessel speed collision avoidance method based on a risk chart modified by measurement uncertainty according to claim 1, characterized in that, Step 1 specifically includes: Step 1.1: Construct the short-time window state sequence of the target ship; Step 1.2: Based on the short-time window state sequence, establish a short-time motion prediction model for the target ship; Step 1.3: Calculate the state prediction residual of the target ship based on the short-time motion prediction model; Step 1.4: Estimate the state uncertainty based on the state prediction residual, obtain the covariance matrix containing the uncertainties of the target ship's position, heading and speed, and perform regularization on the covariance matrix as a measurement uncertainty parameter.

3. The fishing vessel speed collision avoidance method based on a risk chart modified by measurement uncertainty according to claim 1, characterized in that, Step 2 specifically includes: Step 2.1: Construct a discrete grid for the local navigation plane with the ship's current position as the origin; Step 2.2: Calculate the predicted time for the ship to reach each grid point in the discrete grid from its current position; Step 2.3: Predict the virtual state of the ship when it arrives at each grid point; Step 2.4: Predict the state of each target ship at each grid point at the corresponding predicted time; Step 2.5: Calculate the one-sided hazard assessment value based on the virtual state of the ship and the predicted state of the target ship; Step 2.6: Based on the unilateral hazard assessment value, construct the initial spatial collision risk map for each target ship. Step 2.7: When there are multiple target ships, the initial spatial collision risk maps of each single target are fused to obtain the initial spatial collision risk map of multiple targets, which is used as the initial spatial collision risk map.

4. The fishing vessel speed collision avoidance method based on a risk chart revised by measurement uncertainty according to claim 1, characterized by, Step 3 specifically includes: Step 3.1: Extract the uncertainty variances of the target ship's position, heading, and speed from the measurement uncertainty parameters; Step 3.2: Propagate the state uncertainty to the planar position uncertainty to obtain the planar position uncertainty matrix; Step 3.3: Construct the kernel function for the impact of uncertainty based on the planar position uncertainty matrix; Step 3.4: Calculate the uncertainty correction gain based on the planar position uncertainty matrix; Step 3.5: Use the uncertainty-affected kernel function and uncertainty correction gain to correct the initial space collision risk map, and obtain the single-target corrected space collision risk map; Step 3.6: When there are multiple target ships, fuse the single-target corrected spatial collision risk maps to obtain a multi-target corrected spatial collision risk map, which serves as the corrected spatial collision risk map.

5. The fishing vessel speed collision avoidance method based on a risk chart revised by measurement uncertainty according to claim 1, characterized in that, Step 4 specifically includes: Step 4.1: Set risk thresholds; Step 4.2: Extract the grid points with risk values ​​greater than or equal to the risk threshold from the corrected space collision risk map into a comprehensive high-risk area point set; Step 4.3: Based on the comprehensive high-risk area point set, determine the boundary point set of the comprehensive high-risk area; Step 4.4: Extract contour lines, smooth and close the boundary point set to generate a high-risk profile; Step 4.5: Define the closed region enclosed by the high-risk contour as the high-risk envelope region.

6. The method for collision avoidance of fishing vessel speed obstacles based on a risk map corrected for measurement uncertainty as described in claim 5, characterized in that, In step 4.3, the boundary point set of the comprehensive high-risk area is determined using the four-neighbor method.

7. The method for collision avoidance of fishing vessel speed obstacles based on a risk map corrected for measurement uncertainty as described in claim 5, characterized in that, In step 4.4, the generated high-risk profile is smoothed using a moving average method.

8. The method for collision avoidance of fishing vessel speed obstacles based on a risk map corrected for measurement uncertainty as described in claim 1, characterized in that, Step 5 specifically includes: Step 5.1: Based on the ship's safety expansion area, geometrically expand the high-risk envelope area to obtain a comprehensive effective high-risk area; Step 5.2: Determine the candidate speed set based on the ship's permissible candidate speed range and candidate heading angle range; Step 5.3: Determine whether the future trajectory corresponding to each candidate speed in the candidate speed set intersects with the comprehensive effective high-risk area, and take the intersecting candidate speeds as dangerous speeds to construct speed obstacle areas; Step 5.4: Remove speed obstacle regions from the candidate speed set to obtain the safe speed region.

9. The method for collision avoidance of fishing vessel speed obstacles based on a risk map corrected for measurement uncertainty as described in claim 1, characterized in that, Step 6 specifically includes: Step 6.1: Construct a set of safe candidate actions based on the safe speed range; Step 6.2: Construct a motion evaluation function that simultaneously considers the cost of navigation target deviation and the cost of motion smoothness; Step 6.3: Select the candidate action that minimizes the action evaluation function value from the set of safe candidate actions as the optimal collision avoidance action; Step 6.4: Convert the optimal collision avoidance maneuver into heading and speed control commands, and output them for execution after limiting the amplitude.

10. The method for collision avoidance of fishing vessel speed obstacles based on a risk map corrected for measurement uncertainty according to claim 1, characterized in that, In step 6, when a collision risk is detected in the reference direction of the ship toward the target point, the desired avoidance command is calculated only once when the risk first appears. In subsequent control cycles, the ship smoothly approaches the desired avoidance command under the constraints of the rate of change of speed and the rate of change of heading until the risk is eliminated and the ship resumes sailing toward the target point.

Images