Unmanned ship driving path planning method based on automatic search obstacle avoidance

Abstract

The invention relates to the technical field of unmanned vehicle path automatic planning, in particular to an unmanned ship driving path planning method based on automatic search obstacle avoidance, which comprises the following steps: S1, dividing a search map into an obstacle area and a driving area, and determining unmanned ship driving path nodes and a directed graph set; s2, determining unmanned ship driving path tangency points by using a tangency point search function, and setting a peer node principle and a non-child node principle to filter and screen the driving path tangency points; s3, calculating the cost of each driving path by using a heuristic function, and constructing an unmanned ship driving path according to a probability selection principle; and S4, repeating the steps, and comparing, analyzing and determining the optimal driving path of the unmanned ship. The invention provides the unmanned ship driving path planning method capable of automatically searching and avoiding obstacles, the sunlight irradiation principle is simulated, the convergence speed of the algorithm is improved while the shortest path planning is realized, the method has obvious superiority compared with the traditional method, and the feasibility and practicability of the method are proved through comparison simulation experiments.

Description

Technical Field

[0001] The present invention belongs to the technical field of automatic path planning for unmanned vehicles, and in particular relates to a method for planning the path of an unmanned ship based on automatic search and obstacle avoidance. Background Art

[0002] As the top layer of the decision-making and control framework of unmanned vehicles, path planning needs to comprehensively consider the overall environmental information to determine the most effective path and send instructions to the bottom-level control module. Since unmanned vehicles have limited energy and operating power, and need to pay attention to obstacle avoidance at all times during operation, planning the shortest obstacle avoidance path is an important part of the top-level control.

[0003] As a sampling-based path planning algorithm, the Rapid Exploration Random Tree (RRT) is widely used in path planning because of its deterministic search path and strong environmental adaptability. However, the RRT algorithm needs to discretize the space in which it is located during planning. Too few sampling times will affect the performance of the generated path, and too many sampling times will increase the amount of calculation for path planning, thereby reducing real-time performance. Although subsequent algorithms such as the Rapid Exploration Random Tree Star Algorithm (RRT*) and the Improved Rapid Exploration Random Tree (F-RRT) have made certain improvements in path optimization and accelerated convergence speed, the improved optimization algorithm for RRT is still being iterated. Summary of the invention

[0004] In view of the shortcomings of the prior art, the present invention provides a method for unmanned ship driving path planning based on automatic search and obstacle avoidance, which simulates the principle of sunlight exposure to design and plan the driving path, realizes the shortest path planning, and greatly improves the convergence speed of the algorithm, which has obvious advantages over traditional methods.

[0005] To achieve the above object, the present invention discloses the following technical solution:

[0006] A method for planning a driving path of an unmanned ship based on automatic search and obstacle avoidance, comprising:

[0007] S1: Divide the search map into obstacle areas and driving areas, and determine the nodes and directed atlas of the unmanned ship's driving path;

[0008] S11: Divide the entire search area into an obstacle area Xobs and a driving area Xfree, and set the starting node X of the driving path start and the target node X of the driving path goal ;

[0009] S12: Set a directed graph set G = (V, E), where V is the starting node X of the driving path. start and the target node X of the driving path goal All the travel path nodes between X sunis a set of nodes, E is a set of connection relationships between nodes;

[0010] S13: Set the unmanned ship dynamic update data set Openset to record all the unmanned ship alternative driving path nodes X sun The location of the unmanned ship, Pathset records all the alternative driving paths of the unmanned ship;

[0011] S2: Use the tangent point search function to determine the tangent point of the unmanned ship's driving path, and set the same-level node principle and non-child node principle to filter and screen the tangent points of the driving path;

[0012] S21: Set the starting node X of the unmanned boat's driving path in step S1 start The first travel path node X sun , use the tangent point search function to search for the tangent point X of the unmanned ship's planned driving path pt ;

[0013] S22: Set the same-level node principle and non-child node principle to filter and screen the tangent points of the driving path searched in S21, and save the unmanned ship dynamic update data set Openset;

[0014] S3: Set the node X on the unmanned ship's driving path sun Tangent point X with each driving path pt Connect and set up a heuristic function to calculate the cost of each driving path, and determine the next driving path node X based on the probability selection principle sun , construct the unmanned ship driving path and save all the unmanned ship alternative driving paths Pathset;

[0015] S31: The cost function of each unmanned ship's driving path is calculated based on the heuristic function:

[0016] Cost(X)=L total (G,X start ,X)+f(X,X goal );

[0017] Among them, Cost(X) is the cost function of the driving path; L total (G,X start ,X) is the connection function between the node of the driving path and the tangent point of the driving path; f(X,X goal ) is a heuristic function; X is a driving path node dataset; G is a driving path directed graph set function; X start is the starting node of the driving path; X goal is the target node of the driving path;

[0018] S32: Determine the node X of the unmanned ship's driving path according to the probability selection principle sun for:

[0019]

[0020] Among them, X sun is the driving path node; X top is the node with the lowest cost in the driving path node data set; Random is the random generation function; RandomChoose is the random selection function; Openset is the dynamically updated data set; ε is the greedy algorithm parameter;

[0021] S4: repeating steps S2 to S3 to determine the optimal driving path of the unmanned ship through comparative analysis;

[0022] Repeat steps S2 to S3 until the target node X of the driving path is reached. goal , connect each driving path node to form a driving path and put it into the alternative driving path data set Pathset as an alternative driving path; when the set number of simulations is reached, sort the driving paths in the alternative driving path data set Pathset according to the value of the driving path cost, and select the driving path with the smallest driving path cost for forward optimization and backward optimization to obtain the optimal driving path of the unmanned ship.

[0023] Preferably, in step S21, a tangent point search function is used to search for a tangent point X of the unmanned ship's planned driving path. pt , specifically:

[0024] S211: Set the constant angle change value θ step , constant length change value Δ step , forward distance L forward and the positive threshold L there , initialize the iteration angle value θ = 0°, set the alternative tangent point set X candidates , when θ<360°, the iterative distance parameter L and the driving path comparison node X are determined as:

[0025]

[0026] Where L is the iterative distance parameter; Δ step is the constant length change value; X is the driving path comparison node; θ is the iteration angle value;

[0027] S212: When the driving path node X sun The obstacle-free area Xobs between the driving path comparison node X, the updated iterative distance parameter L and the driving path comparison node X are:

[0028]

[0029] S213: When the driving path node X sunWhen an obstacle area Xobs appears between nodes X compared with the driving path, update the iterative distance parameter L = L-Δ step , if the iteration angle value θ>0, then:

[0030]

[0031] Among them, L t is the second iteration distance parameter; θ t is the second iteration angle value; L pre is the first iteration distance step; L forward is the second iteration distance step; θ step is the iteration angle step;

[0032] S214: Calculate the updated driving path comparison node X t for:

[0033] X t =X sun +(L t ·cosθ t ,L t sinθ t );

[0034] Among them, X t Compare nodes for the updated driving path;

[0035] Compare the updated driving path to node X t Assign the tangent point X to the unmanned ship's planned driving path pt , saved in the set of alternative tangent points X on the driving path candidates ;

[0036] S215: After the above steps are completed, determine the first iteration distance step length L pre And the iterative angle value θ is:

[0037]

[0038] Preferably, the same-level node principle in step S22 is specifically as follows: when the driving path node X sun If the tangent point can be connected to its parent node through a line segment, then this tangent point is not its child node but a node of the same level and cannot be put into the dynamically updated dataset Openset.

[0039] Preferably, the non-child node principle in step S22 is specifically as follows: when the driving path node X sun If the tangent point can be connected to the node in the dynamically updated dataset Openset through a line segment, then this tangent point is not its child node and cannot be put into the dynamically updated dataset Openset.

[0040] Preferably, the greedy algorithm parameter ε in step S3 is a constant coefficient between 0 and 1, which is specifically set under the inspiration of the greedy strategy in reinforcement learning. A number between 0 and 1 is randomly selected by a computer program and substituted into the greedy algorithm for simulation, and the optimal value is determined by comparing the simulation results.

[0041] Preferably, the forward optimization in step S4 specifically uses a dichotomy method to optimize the driving path, specifically:

[0042] For the candidate driving path dataset Pathset, remove the starting node X of the driving path start and the target node X of the driving path goal The driving path outside is compared with node X, and the upper node of the driving path is taken as X father , the driving path child node is X son ;

[0043] When the upper node X of the driving path father and the travel path child node X son If the straight line connection is barrier-free, then the driving path is compared with the updated position of node X:

[0044] X=(X father +X son ) / 2;

[0045] Among them, X father X is the upper node of the driving path; son is the driving path subnode;

[0046] When the upper node X of the driving path father and the travel path child node X son If the line segment connection between them is blocked, then the following operations are performed: set the driving path comparison node parameter a = X, and the driving path subnode parameter b = X son ;

[0047] When the distance between the driving path comparison node parameter a and the driving path sub-node parameter b is greater than Δ dichotomy , where the forward accuracy threshold Δ dichotomy ; Then the second parameter of the driving path comparison node is:

[0048] m = (a + b) / 2;

[0049] And make a judgment, when the second parameter m of the driving path comparison node is equal to the driving path upper node X father If the straight line connection between them is unobstructed, then a=m, X=a, otherwise b=m, X=a.

[0050] Preferably, the forward accuracy threshold Δdichotomy It is a positive threshold that affects the accuracy of the binary search and is set according to the actual situation. The specific value of the threshold is 1% of the specific length of the total path.

[0051] Preferably, the backward optimization in step S4 is specifically as follows: the starting node X of the driving path of the unmanned ship is start and the target node X of the driving path goal The nodes are swapped and the superior-subordinate relationships of the nodes in the path are swapped, and then the forward optimization process in step S4 is repeated.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] (1) The present invention provides a method for planning the path of an unmanned ship based on automatic search and obstacle avoidance, simulates the principle of sunlight exposure, and redesigns the path-finding method, thereby greatly improving the convergence speed of the algorithm while achieving the shortest path planning, which is significantly superior to traditional methods.

[0054] (2) The method proposed in the present invention can find the shortest driving path required by the unmanned ship very quickly and accurately, while ensuring the smoothness of the driving path and optimizing the calculation time, effectively solving the problem of the unmanned ship finding the shortest driving path during automatic search. Finally, the feasibility of this method is verified through comparative simulation experiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a flow chart of the unmanned boat driving path planning method based on automatic search and obstacle avoidance of the present invention;

[0056] Figure 2 A process diagram of finding a corresponding tangent point after obtaining a determined driving path node according to the present invention;

[0057] Figure 3 An explanatory diagram for setting the forward distance of the present invention;

[0058] Figure 4 A diagram is provided to demonstrate the tangent performance of the tangent point search function of the present invention;

[0059] Figure 5 Two figures are shown to demonstrate the tangent performance of the tangent point search function of the present invention;

[0060] Figure 6 The schematic diagram of the present invention for screening the obtained cutting points;

[0061] Figure 7 The original image before optimization in the back-end optimization process of the present invention;

[0062] Figure 8 A collision test diagram of the back-end optimization process of the present invention;

[0063] Fig. 9 An updated node graph for the backend optimization process of the present invention;

[0064] Fig.10 It is a first backward optimization result diagram of the back-end optimization process of the present invention;

[0065] Fig.11 It is a second forward optimization result diagram of the back-end optimization process of the present invention;

[0066] Fig.12 It is a second backward optimization result diagram of the back-end optimization process of the present invention;

[0067] Fig.13 This is a diagram of the third forward optimization result of the back-end optimization process of the present invention. DETAILED DESCRIPTION

[0068] The exemplary embodiments, features and aspects of the present invention will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise specified.

[0069] The embodiment of the present invention provides a method for planning a driving path of an unmanned ship based on automatic search and obstacle avoidance. Figure 1 As shown in the figure, the search map is divided into obstacle areas and driving areas, and the nodes and directed graphs of the unmanned ship's driving path are determined; the tangent point search function is used to determine the tangent point of the unmanned ship's driving path, and the same-level node principle and non-child node principle are set to filter and screen the tangent points of the driving path; the heuristic function is used to calculate the cost of each driving path, and the unmanned ship's driving path is constructed according to the probability selection principle; the above steps are repeated to compare and analyze to determine the optimal driving path of the unmanned ship; it includes:

[0070] Step S1: Divide the search map into obstacle areas and driving areas, and determine the nodes and directed atlas of the unmanned ship's driving path.

[0071] Step S11: Divide the entire search area into an obstacle area Xobs and a driving area Xfree, and set the starting node X start and the target node X of the driving path goal .

[0072] Step S12: Set a directed graph set G = (V, E), where V is the starting node X of the driving path. start and the target node X of the driving path goal All the travel path nodes between X sun is a set of nodes, and E is the set of connection relationships between nodes.

[0073] Step S13: Set the unmanned ship dynamic update data set Openset to record all the unmanned ship alternative driving path nodes X sun The Pathset records all the alternative driving paths of the unmanned ship.

[0074] Step S2: Use the tangent point search function to determine the tangent points of the unmanned ship's driving path, and set the same-level node principle and non-child node principle to filter and screen the tangent points of the driving path.

[0075] Step S21: Set the starting node X of the unmanned boat's driving path in step S1 start The first travel path node X sun , use the tangent point search function to search for the tangent point X of the unmanned ship's planned driving path pt , specifically:

[0076] Step S211: Setting a constant angle change value θ step , constant length change value Δ step , forward distance L forward and the positive threshold L there , initialize the iteration angle value θ = 0°, set the alternative tangent point set X candidate , when θ<360°, the iterative distance parameter L and the driving path comparison node X are determined as:

[0077]

[0078] Where L is the iterative distance parameter; Δ step is the constant length change value; X is the driving path comparison node; θ is the iteration angle value.

[0079] like Figure 2 The figure shows the process of finding the corresponding tangent point after obtaining the determined driving path node of the present invention; when the ray moves at the edge of the obstacle, if Δ step When a jump-like change occurs, it means that the edge of the obstacle has been found, so the tangent line is obtained. After a short forward distance L forward Then the cut point can be determined.

[0080] like Figure 3 The figure shows the setting explanation diagram of the forward distance of the present invention; the forward distance L is set forward The purpose is to fit the edge of the curve.

[0081] Step S212: When the driving path node X sun The obstacle-free area Xobs between the driving path comparison node X, the updated iterative distance parameter L and the driving path comparison node X are:

[0082]

[0083] Step S213: When the driving path node X sun When an obstacle area Xobs appears between nodes X compared with the driving path, update the iterative distance parameter L = L-Δ step , if the iteration angle value θ>0, then:

[0084]

[0085] Among them, L t is the second iteration distance parameter; θ t is the second iteration angle value; L pre is the first iteration distance step; L forward is the second iteration distance step; θ step is the iteration angle step size.

[0086] Step S214: Calculate the updated driving path comparison node X t for:

[0087] X t =X sun +(L t ·cosθ t ,L t sinθ t );

[0088] Among them, X t Compare nodes for the updated driving path.

[0089] Compare the updated driving path to node X t Assign the tangent point X to the unmanned ship's planned driving path pt , saved in the set of alternative tangent points X on the driving path candidates .

[0090] Step S215: After the above steps are completed, determine the first iteration distance step length L pre And the iterative angle value θ is:

[0091]

[0092] like Figure 4 The figure shows the tangent performance demonstration of the tangent point search function of the present invention; the dense rays starting from the Sun point first begin to explore the edge position of the obstacle.

[0093] like Figure 5 Shown are two figures demonstrating the tangent performance of the tangent point search function of the present invention; after determining the edge of the obstacle, the location of the specific tangent point can be determined.

[0094] Step S22: Set the same-level node principle and non-child node principle to filter and select the tangent points of the driving path searched in S21, and save the unmanned ship dynamic update data set Openset. Figure 6 The schematic diagram of the present invention for selecting the obtained tangent points is shown; the principle of nodes at the same level is specifically: when the driving path node X sun If the tangent point can be connected to its parent node through the line segment, then this tangent point is not its child node but a peer node and cannot be put into the dynamically updated dataset Openset. The non-child node principle is as follows: when the driving path node X sun If the tangent point can be connected to the node in the dynamically updated dataset Openset through a line segment, then this tangent point is not its child node and cannot be put into the dynamically updated dataset Openset.

[0095] Step S3: Set the node X of the unmanned ship's driving path sun Tangent point X with each driving path pt Connect and set up a heuristic function to calculate the cost of each driving path, and determine the next driving path node X based on the probability selection principle sun , build the unmanned ship driving path and save all the unmanned ship alternative driving paths Pathset.

[0096] Step S31: The cost function of each unmanned boat driving path is calculated according to the heuristic function:

[0097] Cost(X)=L total (G,X start ,X)+f(X,X goal )

[0098] Among them, Cost(X) is the cost function of the driving path; L total (G,X start ,X) is the connection function between the node of the driving path and the tangent point of the driving path; f(X,X goal ) is a heuristic function; X is a driving path node dataset; G is a driving path directed graph set function; X start is the starting node of the driving path; X goal It is the destination node of the driving path.

[0099] Step S32: Determine the node X on the unmanned ship's travel path according to the probability selection principle sun for:

[0100]

[0101] Among them, X sun is the driving path node; X topis the node with the lowest cost in the driving path node data set; Random is the random generation function; RandomChoose is the random selection function; Openset is the dynamically updated data set; ε is the greedy algorithm parameter.

[0102] The greedy algorithm parameter ε is a constant coefficient between 0 and 1, which is specifically set under the inspiration of the greedy strategy in reinforcement learning. A number between 0 and 1 is randomly selected by a computer program and substituted into the greedy algorithm for simulation. The optimal value is determined by comparing the simulation results.

[0103] Step S4: Repeat steps S2 to S3 to determine the optimal driving path of the unmanned ship through comparative analysis.

[0104] Repeat steps S2 to S3 until the target node X of the driving path is reached. goal , connect each driving path node to form a driving path and put it into the alternative driving path data set Pathset as an alternative driving path; when the set number of simulations is reached, sort the driving paths in the alternative driving path data set Pathset according to the value of the driving path cost, and select the driving path with the smallest driving path cost for forward optimization and backward optimization to obtain the optimal driving path of the unmanned ship.

[0105] Forward optimization specifically uses the dichotomy method to optimize the driving path, specifically:

[0106] For the candidate driving path dataset Pathset, remove the starting node X of the driving path start and the target node X of the driving path goal The driving path outside is compared with node X, and the upper node of the driving path is taken as X father , the driving path child node is X son .

[0107] When the upper node X of the driving path father and the travel path child node X son If the straight line connection is barrier-free, then the driving path is compared with the updated position of node X:

[0108] X=(X father +X son ) / 2

[0109] Among them, X father X is the upper node of the driving path; son It is the driving path child node.

[0110] When the upper node X of the driving path father and the travel path child node X sonIf the line segment connection between them is blocked, then the following operations are performed: set the driving path comparison node parameter a = X, and the driving path subnode parameter b = X son .

[0111] When the distance between the driving path comparison node parameter a and the driving path sub-node parameter b is greater than Δ dichotomy， The forward accuracy threshold Δ dichotomy is a positive threshold that affects the accuracy of the binary search. It is set according to the actual situation. The specific value of the threshold is 1% of the specific length of the total path. The second parameter of the driving path comparison node is:

[0112] m=(a+b) / 2

[0113] And make a judgment, when the second parameter m of the driving path comparison node is equal to the driving path upper node X father If the straight line connection between them is unobstructed, then a=m, X=a, otherwise b=m, X=a.

[0114] Backward optimization is: change the starting node X of the unmanned ship's driving path start and the target node X of the driving path goal The nodes are swapped and the superior-subordinate relationships of the nodes in the path are swapped, and then the forward optimization process in step S4 is repeated.

[0115] like Figure 7 The figure shows the pre-optimization diagram of the back-end optimization process of the present invention; it can be seen that there is still room for further optimization of the cost of the current path, which can be achieved by moving certain nodes.

[0116] like Figure 8 The collision test diagram of the back-end optimization process of the present invention is shown; first, it is necessary to determine whether there is an obstacle-free relationship between the node and its parent (superior) child node before optimization. It can be seen from the figure that the straight line connection between the node and its parent and child nodes is obstacle-free, and optimization can be performed according to the obstacle-free determination step in the forward optimization process.

[0117] like Fig. 9 The figure shows the updated node diagram of the back-end optimization process of the present invention; after the collision detection is completed, the node position is moved according to the process to complete the update of the node position. After determining that the collision detection is barrier-free, according to the process, the updated position of the node is the midpoint of the line between its parent and child nodes, and the first forward optimization is completed after the updated position.

[0118] like Fig.10 The first backward optimization result diagram of the back-end optimization process of the present invention is shown; after the first forward optimization is completed, the starting point and the target point are flipped to start the first reverse optimization step. startand the target node X of the driving path goal Make a swap and swap the superior-subordinate relationships of each node in the path, then repeat the forward optimization process.

[0119] like Fig.11 The figure shows the result of the second forward optimization of the back-end optimization process of the present invention; after some obvious node positions are updated, the remaining nodes close to the obstacle will gradually fit toward the edge curve of the obstacle during the update process, which will further shorten the path length and reduce the path cost.

[0120] like Fig.12 The figure shows the result of the second backward optimization of the back-end optimization process of the present invention; the positions of some nodes continue to be optimized.

[0121] like Fig.13 The figure shows the third forward optimization result of the back-end optimization process of the present invention. After further optimizing the position of the path nodes, it can be clearly seen that when the path passes through an obstacle, the position of the path nodes is fitted to the edge curve of the obstacle. After calculating and judging that the cost of the path reaches the target value, it can be output as a result.

[0122] The beneficial effects of the present invention are as follows: the present invention provides a method for planning the driving path of an unmanned ship based on automatic search and obstacle avoidance, simulates the principle of sunlight exposure, and redesigns the path-finding method, thereby greatly improving the convergence speed of the algorithm while realizing the shortest path planning, and has obvious advantages over the traditional method; the embodiments can prove that the present method can very quickly and accurately find the shortest driving path required by the unmanned ship, optimize the calculation time while ensuring the smoothness of the driving path, and effectively solve the problem of the unmanned ship finding the shortest driving path during automatic search. Finally, the feasibility and practicality of the present method are verified through further comparative simulation experiments of the embodiments.

[0123] The embodiments described above are only descriptions of the preferred implementation modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A method for unmanned ship driving path planning based on automatic search and obstacle avoidance, characterized in that: It includes: S1: Divide the search map into obstacle areas and driving areas, and determine the nodes and directed atlas of the unmanned ship's driving path; S11: Divide the entire search area into an obstacle area Xobs and a driving area Xfree, and set the starting node X of the driving path start and the target node X of the driving path goal ; S12: Set a directed graph set G = (V, E), where V is the starting node X of the driving path. start and the target node X of the driving path goal All the travel path nodes between X sun is a set of nodes, E is a set of connection relationships between nodes; S13: Set the unmanned ship dynamic update data set Openset to record all the unmanned ship alternative driving path nodes X sun The alternative driving path dataset Pathset records all the alternative driving paths of the unmanned ship; S2: Use the tangent point search function to determine the tangent point of the unmanned ship's driving path, and set the same-level node principle and non-child node principle to filter and screen the tangent points of the driving path; S21: Set the starting node X of the unmanned boat's driving path in step S1 start The first travel path node X sun , use the tangent point search function to search for the tangent point X of the unmanned ship's planned driving path pt ; S22: Set the same-level node principle and non-child node principle to filter and screen the tangent points of the driving path searched in S21, and save the unmanned ship dynamic update data set Openset; S3: Set the node X on the unmanned ship's path sun Tangent point X with each driving path pt Connect and set up a heuristic function to calculate the cost of each driving path, and determine the next driving path node X based on the probability selection principle sun , construct the unmanned ship driving path and save all the unmanned ship alternative driving paths Pathset; S31: The cost function of each unmanned ship's driving path is calculated based on the heuristic function: Cost(X)=L total (G,X start ,X)+f(X,X goal ) Among them, Cost(X) is the cost function of the driving path; L total (G,X start ,X) is the connection function between the node of the driving path and the tangent point of the driving path; f(X,X goal ) is a heuristic function; X is a driving path node dataset; G is a driving path directed graph set function; X start is the starting node of the driving path; X goal is the target node of the driving path; S32: Determine the node X of the unmanned ship's driving path according to the probability selection principle sun for: Among them, X sun is the driving path node; X top is the node with the lowest cost in the driving path node data set; Random is the random generation function; RandomChoose is the random selection function; Openset is the dynamically updated data set; ε is the greedy algorithm parameter; S4: repeating steps S2 to S3 to determine the optimal driving path of the unmanned ship through comparative analysis; Repeat steps S2 to S3 until the target node X of the driving path is reached. goal , connect each driving path node to form a driving path and put it into the alternative driving path data set Pathset as an alternative driving path; when the set number of simulations is reached, sort the driving paths in the alternative driving path data set Pathset according to the value of the driving path cost, and select the driving path with the smallest driving path cost for forward optimization and backward optimization to obtain the optimal driving path of the unmanned ship.

2. The method for unmanned ship driving path planning based on automatic search and obstacle avoidance according to claim 1 is characterized in that: In step S21, a tangent point search function is used to search for the tangent point X of the unmanned ship's planned driving path. pt , specifically: S211: Set the constant angle change value θ step , constant length change value Δ step , forward distance L forward and the positive threshold L there , initialize the iteration angle value θ = 0°, set the alternative tangent point set X candidates , when θ<360°, the iterative distance parameter L and the driving path comparison node X are determined as: Where L is the iterative distance parameter; Δ step is the constant length change value; X is the driving path comparison node; θ is the iteration angle value; S212: When the driving path node X sun The obstacle-free area Xobs between the driving path comparison node X, the updated iterative distance parameter L and the driving path comparison node X are: S213: When the driving path node X sun When an obstacle area Xobs appears between nodes X compared with the driving path, update the iterative distance parameter L = L-Δ step , if the iteration angle value θ>0, then: Among them, L t is the second iteration distance parameter; θ t is the second iteration angle value; L pre is the first iteration distance step; L forward is the second iteration distance step; θ step is the iteration angle step; S214: Calculate the updated driving path comparison node X t for: X t =X sun +(L t ·cosθ t ,L t ·sinθ t ) Among them, X t Compare nodes for the updated driving path; Compare the updated driving path to node X t Assign the tangent point X to the unmanned ship's planned driving path pt , saved in the set of alternative tangent points X on the driving path candidates ; S215: After the above steps are completed, determine the first iteration distance step length L pre And the iterative angle value θ is:

3. The method for unmanned ship driving path planning based on automatic search and obstacle avoidance according to claim 1 is characterized in that: The principle of nodes at the same level in step S22 is as follows: when the node X sun If the tangent point can be connected to its parent node through a line segment, then this tangent point is a node of the same level and cannot be put into the dynamically updated dataset Openset.

4. The unmanned ship driving path planning method based on automatic search and obstacle avoidance according to claim 1 is characterized in that: The non-child node principle in step S22 is as follows: when the driving path node X sun If the tangent point can be connected to the node in the dynamically updated dataset Openset through a line segment, then this tangent point is not its child node and cannot be put into the dynamically updated dataset Openset.

5. The method for unmanned ship driving path planning based on automatic search and obstacle avoidance according to claim 1 is characterized in that: The greedy algorithm parameter ε in step S3 is a constant coefficient between 0 and 1, which is specifically set under the inspiration of the greedy strategy in reinforcement learning. A number between 0 and 1 is randomly selected by a computer program and substituted into the greedy algorithm for simulation, and the optimal value is determined by comparing the simulation results.

6. The unmanned ship driving path planning method based on automatic search and obstacle avoidance according to claim 1 is characterized in that: The forward optimization in step S4 specifically uses the dichotomy method to optimize the driving path, specifically: For the candidate driving path dataset Pathset, remove the starting node X of the driving path start and the target node X of the driving path goal The driving path outside is compared with node X, and the upper node of the driving path is taken as X father , the driving path child node is X son ; When the upper node X of the driving path father and the travel path child node X son If the straight line connection is barrier-free, then the driving path is compared with the updated position of node X: X=(X father +X son ) / 2 Among them, X father X is the upper node of the driving path; son is the driving path subnode; When the upper node X of the driving path father and the travel path child node X son If the line segment connection between them is blocked, then the following operations are performed: set the driving path comparison node parameter a = X, and the driving path subnode parameter b = X son ; When the distance between the driving path comparison node parameter a and the driving path sub-node parameter b is greater than Δ dichotomy， The forward accuracy threshold Δ dichotomy ; Then the second parameter of the driving path comparison node is: m=(a+b) / 2 And make a judgment, when the second parameter m of the driving path comparison node is equal to the driving path upper node X father If the straight line connection between them is unobstructed, then a=m, X=a, otherwise b=m, X=a.

7. The method for unmanned ship driving path planning based on automatic search and obstacle avoidance according to claim 6 is characterized in that: The forward accuracy threshold Δ dichotomy It is a positive threshold that affects the accuracy of the binary search and is set according to the actual situation. The specific value of the threshold is 1% of the specific length of the total path.

8. The unmanned ship driving path planning method based on automatic search and obstacle avoidance according to claim 1 is characterized in that: The backward optimization in step S4 is specifically as follows: start and the target node X of the driving path goal The nodes are swapped and the superior-subordinate relationships of the nodes in the path are swapped, and then the forward optimization process in step S4 is repeated.

Images