An unmanned ship multi-constraint local path planning method based on visual information

Abstract

This invention provides a multi-constraint local path planning method for unmanned surface vessels (USVs) based on visual information, relating to the field of USV motion planning technology. The method includes the following steps: modeling the surrounding environment of the USV based on depth image information to obtain a 3D occupancy grid map; using the 3D occupancy grid map as the map information for path planning, searching for path points on the 3D occupancy grid map using a Durbins curve to obtain path points; obtaining a reference centerline; dynamically expanding the path points in the curve normal direction to determine the left and right boundaries of narrow channels, obtaining lateral distance constraints; and solving a quadratic programming problem involving lateral distance, yaw angle, curvature, rate of change of curvature, and relaxation variables to obtain the optimal path that the USV can ultimately traverse. This invention adds shoreline constraints to the path point search for further optimization, improving the safety of the USV's navigation path.

Description

Technical Field

[0001] This invention relates to the field of unmanned surface vessel (USV) motion planning technology, and more particularly to a multi-constraint local path planning method for USVs based on visual information. Background Technology

[0002] Unmanned surface vessels (USVs), characterized by their high speed and maneuverability, are widely used not only in marine environmental monitoring and search and rescue but also play a vital role in the military field, becoming an important tool for exploring the marine environment and developing marine resources. USVs can autonomously navigate complex environments without human intervention, after determining their course, achieving obstacle and collision avoidance, thus helping people better accomplish a variety of complex tasks in the marine environment.

[0003] Narrow waterways are widely distributed in rivers both domestically and internationally, as well as narrow waterways at sea, and constitute challenging sections for unmanned surface vessels (USVs). Compared to open waters, the environment in narrow waterways is more complex, constrained by factors such as shorelines, and the navigation range is smaller. Therefore, this places higher demands on path planning for USVs in narrow waterways. Path planning techniques used in route planning have received significant attention in USV research in recent years. These techniques generally include three parts: environment modeling, path search, and path smoothing. Environment modeling is achieved through perception of the surrounding environment. Based on the modeled environment, a path search is performed to obtain a path that avoids obstacles. Finally, an optimization problem is constructed to optimize the path, resulting in the final path for the USV's movement. This is extremely important for USV navigation in narrow waterways.

[0004] Currently available path planning methods for unmanned surface vessels (USVs) mostly rely on satellite or high-resolution maps. These methods suffer from several problems: First, while A* algorithm and fast search random tree algorithm are widely used, combining the advantages of breadth-first search and depth-first search, they can solve route planning problems in narrow waterways. However, due to the heuristic nature of these algorithms, the search route may approach the shoreline and obstacles, compromising safety. Second, these algorithms depend on grid maps constructed from electronic nautical charts, which are highly limited by the grid structure and cannot update nearby obstacle information in real time. This makes them insufficient for smoothing routes or optimizing path movement, failing to meet the needs of USV motion planning. Third, they lack further map optimization on the existing two-dimensional grid map to enrich and improve map information. Furthermore, grid information alone is insufficient for complex trajectory optimization, thus compromising the safety of the planned USV path. Summary of the Invention

[0005] In view of this, the purpose of this invention is to propose a multi-constraint local path planning method for unmanned surface vessels (USVs) based on visual information. Based on binocular visual information, the method performs modeling in the local environment, and through the search of front-end path points, path smoothing, and path optimization, it finally generates a path plan for the USV's navigation path, thereby solving the technical problem of poor safety in existing narrow path planning methods for USVs.

[0006] The technical means employed in this invention are as follows:

[0007] A method for multi-constraint local path planning of unmanned surface vessels based on visual information includes the following steps:

[0008] Depth image information is acquired using a monocular depth camera. The current position and attitude of the unmanned surface vessel are obtained by using a synchronous positioning and real-time mapping method on the depth image information. The surrounding environment of the unmanned surface vessel is modeled based on the depth image information to obtain a three-dimensional occupancy grid map.

[0009] Using the three-dimensional occupied grid map as the map information relied upon for path planning, the Dubins curve is used to search for path points on the three-dimensional occupied grid map to obtain path points.

[0010] A reference centerline is obtained by fitting a cubic B-spline curve that does not pass through the control points to the path points.

[0011] A quadratic programming problem is constructed on the reference centerline with respect to the lateral distance, the first derivative of the curve, the second derivative of the curve, and the third derivative of the curve to obtain the smoothed curve;

[0012] The smoothed curve is discretized based on its curvature to obtain discrete sampling points;

[0013] Dynamic expansion of path points along the curve normal direction determines the left and right boundaries of the narrow river channel, thus obtaining lateral distance constraints.

[0014] A quadratic programming problem is constructed on the lateral distance constraint obtained by combining the discrete sampling points with the expansion. The problem is then solved to obtain the optimal path that the unmanned surface vessel can finally pass through.

[0015] Furthermore, establishing a 3D occupied grid map includes the following steps:

[0016] The depth image information is processed into grayscale and edge detection algorithms are used for edge extraction. Pixels are read one by one, and non-zero pixels are converted into grids. The map is then scaled to create a 3D map of the narrow river channel. A 3D occupied grid map is constructed based on the location of the unmanned surface vessel (USV). An occupied grid is represented by 1, and an empty grid by 0. The posterior probability of the depth image information observed by the monocular depth camera on the USV for the tth time represents the occupancy status of the grid. Let g be the occupancy status of the grid. i The t-th observation of the monocular depth camera is recorded as l. t (g i The cumulative observations of the first t-1 times are denoted as l. t-1 (g i The formula for a 3D grid map is as follows:

[0017]

[0018] p(z t |g i () is a sensor model that updates in real time based on the information obtained by the sensor each time. t (g i ); when l t (g i When the cumulative value of ) exceeds 1, it indicates that the grid is occupied; when the cumulative value is less than 0, it indicates that the grid is idle.

[0019] Furthermore, the search for waypoints on a 3D occupied raster map using the Dubins curve includes:

[0020] The Dubins curve includes three basic unmanned surface vessel (USV) maneuvers: left turn (L), right turn (R), and straight ahead (S). For any given starting point and ending point, there are six corresponding curves to choose from, as shown in the following set:

[0021] {LRL,LSL,LSR,RLR,RSR,RSL}

[0022] Calculate the distance corresponding to each generated curve, select the curve with the smallest distance between two given points as the Dubins curve between the two points; discretize the generated trajectory and finally obtain the required path points.

[0023] Furthermore, the form of the quadratic programming problem in terms of lateral distance, the first derivative of the curve, the second derivative of the curve, and the third derivative of the curve is as follows:

[0024]

[0025] in, The distance from the discrete points of the B-spline curve. The first derivative of the curve representing velocity, To represent the second derivative of the curve representing acceleration, To represent the third derivative of the curve for jerk, (x ni ,y ni ) represents the position variable, l' represents the velocity variable, l″ represents the acceleration variable, l'" represents the jerk variable, and ω represents the acceleration variable. l ,ω l' ,ω l" ,ω l"' This is the weight value.

[0026] Further, determining the left and right boundaries of the narrow channel includes: expanding left and right along the normal direction based on the sampled path points; determining the distance from the center of each grid cell to the nearest obstacle using the Euclidean distance algorithm; calculating the distance to the nearest obstacle at any expanded point using the trilinear interpolation algorithm; considering the distance threshold to the obstacle, calculating the leftmost and rightmost boundaries closest to the obstacle; if the distance between the sampled path points and the obstacle is less than the threshold before expansion, the path points need to be resampled, and then the left and right boundaries need to be calculated.

[0027] Furthermore, the quadratic programming problem concerning lateral distance, bow angle, curvature, rate of change of curvature, and slack variables is constructed as follows:

[0028]

[0029] in, This represents the change in lateral distance between the path point and the sampled path points. The change in bow roll angle of the unmanned surface vessel. The curvature of the path points. The rate of change of curvature at the path points. These are slack variables.

[0030] The present invention also provides a storage medium comprising a stored program, wherein, when the program is executed, any of the above-described unmanned surface vessel multi-constraint local path planning methods based on visual information are performed.

[0031] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes any of the above-described visual information-based unmanned surface vessel multi-constraint local path planning methods through the computer program.

[0032] Compared with the prior art, the present invention has the following advantages:

[0033] First, compared with the existing A* algorithm and fast search random tree algorithm for unmanned surface vessel (USV) path planning, this invention adds shoreline constraints to the path point search for further optimization, thereby improving the safety of USV navigation paths.

[0034] Secondly, compared with existing unmanned surface vessels (USVs) that rely on grid information in electronic nautical charts for path planning, this invention uses point cloud information from real-time visual feedback to construct a grid map. The grid size can be adjusted to further improve the efficiency of waypoint search. At the same time, the map range is relatively small, which can well meet the requirements of smooth USV routes.

[0035] Third, compared with the existing unmanned surface vessels that rely solely on grid maps for planning, this invention, based on the occupation of grids, uses linear interpolation to further calculate the distance between path points in the occupied grid map and the nearest obstacle, providing more information for subsequent optimization of path points and more conveniently solving the problem of obstacle avoidance distance calculation in narrow river conditions. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 This is a flowchart of the present invention.

[0038] Figure 2 This is a schematic diagram of a three-dimensional map of the narrow river channel constructed according to the present invention.

[0039] Figure 3 This is a schematic diagram of an algorithm for path point search along the RSL direction based on the Durbins curve.

[0040] Figure 4 This is a schematic diagram of an algorithm for path point search along the LSL direction based on the Durbins curve.

[0041] Figure 5 This is a schematic diagram of cubic B-spline curve fitting for the path points obtained in this invention.

[0042] Figure 6 This is a schematic diagram illustrating how the left and right boundary constraints are obtained based on smooth path points according to the present invention.

[0043] Figure 7 This is a schematic diagram illustrating the autonomous navigation of a single unmanned surface vessel in a narrow river environment using the optimal path obtained through quadratic programming.

[0044] Figure 8 This is a schematic diagram illustrating the path planning of a single unmanned surface vessel in a simulated narrow river environment according to the present invention. Detailed Implementation

[0045] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0046] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0047] This invention provides a multi-constraint local path planning method for unmanned surface vessels based on visual information, comprising the following steps:

[0048] A. Three-dimensional simulation of narrow river channel environment and establishment of three-dimensional occupied grid map

[0049] The images from the simulation environment are processed into grayscale and edge detection algorithms are used for edge extraction. Pixels are read one by one, and non-zero pixels are converted into grids. The map is then scaled to create a 3D map of the narrow river channel. A 3D occupied grid map is constructed based on the location of the unmanned surface vessel (USV). Occupied grids are represented by 1, and empty grids by 0. The posterior probability of the depth image information observed by the monocular depth camera on the USV on the t-th time represents the occupancy status of the grid. Let the occupancy status of grid g be... i The t-th observation of the monocular depth camera is recorded as l. t (g i The cumulative observations of the first t-1 times are denoted as l. t-1 (g i Then, we can obtain the following formula:

[0050]

[0051] p(z t |g i () is a sensor model that updates in real time based on the information obtained by the sensor each time.t (g i When l t (g i When the cumulative value of ) exceeds 1, it indicates that the grid is occupied; when the cumulative value is less than 0, it indicates that the grid is idle.

[0052] B. Path point search based on Durbins curve

[0053] The kinematic model of the unmanned surface vessel is shown below:

[0054]

[0055] Where r is the bow angular velocity of the unmanned surface vessel (USV), υ is the longitudinal velocity of the USV, and θ is the bow angle of the USV. Discretizing the above kinematic equations and taking the sampling time as dt, let the current position coordinates and bow angle of the USV be (x0, y0, θ0), and the position coordinates and bow angle after the motion be (x...). t y t θ t The formula for an unmanned surface vessel moving from one state to another is shown below:

[0056]

[0057] The Dubins curve encompasses three basic maneuvers of unmanned surface vessels (USVs): left turn (L), right turn (R), and straight ahead (S). For any given starting and ending point, there are six possible curves to choose from, as shown in the following set:

[0058] {LRL,LSL,LSR,RLR,RSR,RSL}(4)

[0059] Calculate the distance corresponding to each generated curve, and select the curve with the smallest distance between two given points as the Dubins curve between the two points. Discretize and sample the generated trajectory to finally obtain the required path points.

[0060] C. Fitting the path points with a cubic B-spline curve without controlling the points.

[0061] Given that four discrete points in a plane can determine a cubic B-spline curve, the path point search based on the Dubins curve is required to generate at least four path points. Assuming there are n path points, n-3 consecutive B-spline curves can be determined. Let the four discrete points of the i-th B-spline curve be denoted as (x...). i ,y i ),(x i+1 ,y i+1 ),(x i+2 ,y i+2 ),(x i+3 ,y i+3Set the curve parameter t to a range of (0,1). For the x-axis and y-axis, set the x-axis curve component coefficient variable to (a). i0 ,a i1 ,a i2 ,a i3 The coefficient variables of the y-axis curve components are (b) i0 ,b i1 ,b i2 ,b i3 The parametric equations of the curves corresponding to the x-axis and y-axis are x i (t),y i (t) is shown below:

[0062]

[0063] The corresponding coefficient variables are expressed in the following forms:

[0064]

[0065] By taking the reciprocal c of the i-th segment of the B-spline curve, t is divided into c equal parts to obtain discrete points and the distance between discrete points.

[0066] D. Construct a quadratic programming problem concerning curve smoothness.

[0067] Having obtained the discrete cubic B-spline curve points mentioned above, a quadratic programming problem is constructed based on these discrete points, relating the distance from the reference center, the first derivative of the curve, the second derivative of the curve, and the third derivative of the curve. The form of the constructed problem is shown below:

[0068]

[0069] This represents the distance from the discrete points of the B-spline curve. It represents the first derivative of the curve and is used to represent velocity. It represents the second derivative of the curve and is used to represent acceleration. This represents the third derivative of the curve, used to represent jerk. The position variable (x) needs optimization. ni ,y ni ), velocity variable l', acceleration variable l″, accelerometer variable l'", ω l ,ω l' ,ω l" ,ω l"' These represent the corresponding weight values. Since the curve uses distance as a parameter, the following relationship can be obtained:

[0070]

[0071] Substituting the above formula into the optimization problem, after transformation, we obtain the following relationship:

[0072]

[0073] Construct constraints for the x-axis and y-axis according to formula (8), and assume the curvature of the i-th point is κ. i , used to replace the second derivative at that point. The constraints of this quadratic programming problem are as follows:

[0074]

[0075] Where i∈[0,n], the smoothed path points are obtained by solving the above quadratic programming problem. The path points are then resampled according to the different curvatures, and the sampling formula is as follows:

[0076]

[0077] Let the maximum curvature be κ. ub The minimum curvature is κ lb The maximum sampling distance is s. ub The minimum sampling distance is s lb κ is determined based on the different curvatures at the path points. i Therefore, the sampling distance Δs is determined. i To ensure that pathpoints are sampled more densely when curvature is high and less densely when curvature is low, the sampling is performed horizontally and vertically along the normal direction based on the sampled pathpoints. The distance from each grid center to the nearest obstacle is determined using the Euclidean distance algorithm, and the distance to the nearest obstacle at any expanded point is calculated using trilinear interpolation. Considering a distance threshold to obstacles, the leftmost and rightmost boundaries closest to obstacles are calculated. If, before expansion, the distance between a sampled pathpoint and an obstacle is less than the threshold, pathpoint sampling needs to be repeated, and then the left and right boundaries calculated.

[0078] E. Construct a quadratic programming problem for generating safe paths.

[0079] E1. Construction of Quadratic Programming Problems

[0080] The problem is constructed in the following form:

[0081]

[0082] This represents the change in lateral distance between the path point and the sampled path points. The inner part represents the change in the bow roll angle of the unmanned surface vessel. The curvature of the path points. The rate of change of curvature at the path points. The ω is a slack variable. Slack variables are used to ensure that the quadratic programming problem has a solution and to accelerate convergence.pq ,ω θ ,ω c ,ω cr ,ω s These are the corresponding weighting coefficients.

[0083] E2, Unmanned Surface Vehicle Collision Detection

[0084] The unmanned surface vessel (USV) is covered by four circumscribed circles. Collision detection of the USV's rectangular outline is transformed into collision detection of circles. Let the coordinates of the centers of the four circumscribed circles covering the USV at the i-th path point be (c...). i1x ,c i1y ),(c i2x ,c i2y ),(c i3x ,c i3y ),(c i4x ,c i4y The radius is set manually, here we set it to r, and the distance from the center of the rectangle is d. i1 ,d i2 ,d i3 ,d i4 The unmanned surface vessel's bow angle is θ. i The coordinates of the path point are (x i ,y i The expression for the circumcenter is as follows:

[0085]

[0086] The distance from the center of each circle to the nearest obstacle is calculated using trilinear interpolation and compared with the radius of the circle to achieve collision detection for the unmanned surface vessel.

[0087] Construction of E3, boundary constraints, and curvature constraints, etc.

[0088] The following constraints are constructed for the quadratic programming problem of generating safe paths, with the constraint on the change in lateral distance given by the formula below:

[0089] Δl i +Δs i Δθ i -Δl i+1 =0 (14)

[0090] Where i∈[0,n-2], Δs i Let be the distance between the i-th path point and the (i+1)-th path point. The constraint on curvature is given by the following formula:

[0091]

[0092] Where i∈[0,n-2], κ lbκ represents the minimum curvature. ub The maximum value of curvature represents the curvature change satisfying the above equation constraint, while the curvature value is within a set range. In the step of expanding left and right to obtain the left and right boundaries closest to the obstacle after smoothing the path, the left boundary obtained from the i-th path point relative to the four centers of the collision detection is l. i0 ,l i1 ,l i2 ,l i3 The right boundary is r i0 ,r i1 ,r i2 ,r i3 The collision constraints obtained for the left and right boundaries are as follows:

[0093]

[0094] Where i∈[0,n-1], the optimized path points and the bow angle of the unmanned surface vessel corresponding to each path point can be obtained by solving the quadratic programming problem proposed by formula (12) and the constraints listed above.

[0095] This invention can be summarized in the following steps:

[0096] The first step is to construct a scene of the narrow river channel based on the existing map, using image information to create a scaled-down representation, such as... Figure 2 As shown, based on the provided map information, a path point search is performed using the Durbins curve, as follows: Figure 3 and Figure 4 As shown, the unmanned surface vessel (USV) is connected between a given starting point and a given ending point using a Durbins curve. After obtaining the path points, a B-spline curve is used for fitting, as shown... Figure 5 As shown.

[0097] The second step, based on the obtained B-spline path, constructs a quadratic programming problem with respect to the lateral distance, the first derivative, the second derivative, and the third derivative of the curve, using the sampling points to obtain a smooth curve. Then, the path points are expanded along the normal direction. A distance threshold is set to obtain the left and right boundary points, such as... Figure 6 As shown. Finally, based on the left and right boundary points, a quadratic programming problem is constructed with respect to the changes in lateral distance, bow angle, curvature, and slack variables. Solving this problem yields a safe path. The implementation process and the optimized path during the motion are shown below. Figure 7 and Figure 8 As shown.

[0098] When planning the path of the unmanned surface vessel (USV), the initial pose is obtained by visual mapping. The target point is selected by clicking on the simulation platform of the constructed narrow river channel. The USV moves according to the optimized path points and finally moves to the destination.

[0099] The present invention also provides a storage medium comprising a stored program, wherein, when the program is executed, the above-described unmanned surface vessel multi-constraint local path planning method based on visual information is performed.

[0100] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the above-described unmanned surface vessel multi-constraint local path planning method based on visual information through the computer program.

[0101] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0102] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0103] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0104] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0105] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0106] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0107] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multi-constraint local path planning method for unmanned surface vessels based on visual information, characterized in that, Includes the following steps: Depth image information is acquired using a monocular depth camera. The current position and attitude of the unmanned surface vessel are obtained by using a synchronous positioning and real-time mapping method on the depth image information. The surrounding environment of the unmanned surface vessel is modeled based on the depth image information to obtain a three-dimensional occupancy grid map. Using the three-dimensional occupied grid map as the map information relied upon for path planning, the Dubins curve is used to search for path points on the three-dimensional occupied grid map to obtain path points. A reference centerline is obtained by fitting a cubic B-spline curve that does not pass through the control points to the path points. A quadratic programming problem is constructed on the reference centerline with respect to the lateral distance, the first derivative of the curve, the second derivative of the curve, and the third derivative of the curve to obtain the smoothed curve; The smoothed curve is discretized based on its curvature to obtain discrete sampling points; Dynamic expansion of path points along the curve normal direction determines the left and right boundaries of the narrow river channel, thus obtaining lateral distance constraints. A quadratic programming problem is constructed on the lateral distance constraint obtained by combining the discrete sampling points with the expansion. The problem is then solved to obtain the optimal path that the unmanned surface vessel can finally pass through.

2. The unmanned surface vessel multi-constraint local path planning method based on visual information according to claim 1, characterized in that, Creating a 3D occupied grid map involves the following steps: The depth image information is processed into grayscale and edge detection algorithms are used for edge extraction. Pixels are read one by one, and non-zero pixels are converted into grids. The map is then scaled to create a 3D map of the narrow river channel. A 3D occupied grid map is constructed based on the location of the unmanned surface vessel (USV). An occupied grid is represented by 1, and an empty grid by 0. The posterior probability of the depth image information observed by the monocular depth camera on the USV for the tth time represents the occupancy status of the grid. Let g be the occupancy status of the grid. i The t-th observation of the monocular depth camera is recorded as l. t (g i The cumulative observations of the first t-1 times are denoted as l. t-1 (g i The formula for a 3D grid map is as follows: p(z t |g i () is a sensor model that updates in real time based on the information obtained by the sensor each time. t (g i ); when l t (g i When the cumulative value of ) exceeds 1, it indicates that the grid is occupied; when the cumulative value is less than 0, it indicates that the grid is idle.

3. The unmanned surface vessel multi-constraint local path planning method based on visual information according to claim 1, characterized in that, The search for path points in a 3D occupied raster map using the Dubins curve includes: The Dubins curve includes three basic unmanned surface vessel (USV) maneuvers: left turn (L), right turn (R), and straight ahead (S). For any given starting point and ending point, there are six corresponding curves to choose from, as shown in the following set: {LRL,LSL,LSR,RLR,RSR,RSL} Calculate the distance corresponding to each generated curve, select the curve with the smallest distance between two given points as the Dubins curve between the two points; discretize the generated trajectory and finally obtain the required path points.

4. The unmanned surface vessel multi-constraint local path planning method based on visual information according to claim 1, characterized in that, The quadratic programming problem in terms of lateral distance, first derivative of the curve, second derivative of the curve, and third derivative of the curve takes the form shown below: in, The distance from the discrete points of the B-spline curve. The first derivative of the curve representing velocity, To represent the second derivative of the curve representing acceleration, To represent the third derivative of the curve for jerk, (x ni ,y ni ) represents the position variable, l' represents the velocity variable, l” represents the acceleration variable, l'" represents the jerk variable, and ω represents the acceleration variable. l ,ω l' ,ω l" ,ω l"' This represents the weight value.

5. The unmanned surface vessel multi-constraint local path planning method based on visual information according to claim 1, characterized in that, Determining the left and right boundaries of the narrow channel includes: expanding left and right along the normal direction based on the sampled path points; determining the distance from the center of each grid cell to the nearest obstacle using the Euclidean distance algorithm; calculating the distance to the nearest obstacle at any expanded point using the trilinear interpolation algorithm; considering the distance threshold to the obstacle, calculating the leftmost and rightmost boundaries closest to the obstacle; if the distance between the sampled path points and the obstacle is less than the threshold before expansion, the path points need to be resampled, and then the left and right boundaries need to be calculated.

6. The unmanned surface vessel multi-constraint local path planning method based on visual information according to claim 1, characterized in that, The quadratic programming problem with respect to lateral distance, bow angle, curvature, rate of change of curvature, and slack variables takes the following form: in, This represents the change in lateral distance between the path point and the sampled path points. The change in bow roll angle of the unmanned surface vessel. The curvature of the path points. The rate of change of curvature at the path points. These are slack variables.

7. A storage medium, characterized in that, The storage medium includes a stored program, wherein when the program is executed, it performs the visual information-based unmanned surface vessel multi-constraint local path planning method as described in any one of claims 1 to 6.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The processor executes the visual information-based unmanned surface vessel multi-constraint local path planning method according to any one of claims 1 to 6 through the computer program.

Images