CN113848897A - Method, system and related product for path planning for unmanned surface vessels - Google Patents

Abstract

The present disclosure relates to a method for path planning for an unmanned surface craft, the method is executed by an edge server, and includes: receiving a path planning request sent from a target unmanned surface craft; determine the final path of the target unmanned surface craft within the coverage area of the edge server or outside the coverage area, wherein when the path planning request is outside the coverage area of the edge server, send a path planning request to the cloud server to obtain the cloud The server determines the final path based on the preliminary path planning returned by the path planning request; and determines the final path based on the preliminary path planning and the path cache in the edge server; wherein when the path planning request is within the coverage area of the edge server, based on the path planning request and The path cache in the edge server determines the final path so that the target USV will follow the final path. With the solution of the present disclosure, the response speed of the path request and the path planning efficiency can be improved.

Description

Methods, systems and related products for path planning of unmanned surface craft

technical field

The present disclosure generally relates to the technical field of unmanned surface vehicles. More specifically, the present disclosure relates to a method, system, and computer-readable storage medium for path planning for an unmanned surface craft.

Background technique

With the development of advanced robotics and automation technology, Unmanned Surface Vessel (“USV”), as an important mobile robot, is widely used in ocean sampling, rescue, environmental monitoring, military reconnaissance and other fields. In the aforementioned application fields, a common problem that needs to be solved is to plan a feasible path according to requirements in an uncertain environment.

Currently, a unified scheduling method is usually used to plan paths for USVs, that is, to process all data and computing requests of USVs through the cloud. However, due to the instability of the Internet, data packet loss and transmission delay caused by network fluctuations are prone to occur during multi-hop transmission of data, resulting in uncertain network delays between the cloud and the USV. During the path planning process, the uncertain delay will directly affect the real-time performance and obstacle avoidance performance of the USV's path planning. In addition, the computational cost of querying the optimal path on the cloud server will increase rapidly with the increase of the length and resolution of the target path. As the distance from the target node increases, the computational time of these algorithms increases exponentially.

Edge computing is a new computing architecture that enables computing, storage, and network resources to be integrated at the edge of the network, closer to end users. In order to reduce the transmission delay of path planning queries and reduce the amount of cloud computing, edge computing-based path planning algorithms have been proposed in recent years. That is, the route is sent to an edge server placed on the shore close to the USV. In addition, in order to further improve the efficiency of path planning, related scholars have also proposed some cache-based path planning algorithms. Examples include benefit-driven shortest-path caching algorithms to answer shortest-path queries from source and destination nodes, and cache-based path planning by Caching ("PPC") systems, which utilize partially matched cached paths to answer a new query. However, most cache-based path planning algorithms adopt a point-to-point matching mode, which not only brings huge storage overhead to the server, but also greatly reduces the cache path query hit rate.

SUMMARY OF THE INVENTION

In order to at least partially solve the technical problems mentioned in the background art, the solution of the present disclosure provides a solution for path planning for an unmanned surface craft. Using the solution of the present disclosure, the edge server and the cloud server are used to perform path planning for the unmanned surface vessel, and the final path is determined through the assistance of the path cache, thereby improving the path planning efficiency and improving the response speed of the path request. To this end, the present disclosure provides solutions in the following aspects.

In one aspect, the present disclosure provides a method for path planning for an unmanned surface vehicle, the method being performed by an edge server, and comprising: receiving a path planning request sent from a target unmanned surface vehicle; the path planning request is within or outside the coverage area of the edge server to determine the final path of the target USV, wherein when the path planning request is outside the coverage area of the edge server, the method comprising: sending the path planning request to a cloud server to obtain a preliminary path planning returned by the cloud server based on the path planning request; and determining a final path based on the preliminary path planning and a path cache in the edge server; Wherein, when the path planning request is within the coverage area of the edge server, the method includes: determining a final path based on the path planning request and a path cache in the edge server, so that the target unmanned surface vehicle Follow the final path.

In one embodiment, wherein the preliminary path planning includes entry points and exit points passing through a coverage area of a corresponding edge server, wherein determining a final path based on the preliminary path planning and a path cache in the edge server includes querying a path whether there is a path in the cache containing the entry point and exit point; and determining the final path based on the query result.

In another embodiment, the determining of the final path based on the query result includes: when there is a path including the entry point and the exit point in the path cache, the path cache passes through the entry point and the exit point. The path of the point is used as the final path; or when the path cache does not contain a path including the entry point and the exit point, the final path is determined according to the grid information of the entry point and the exit point in the path cache. .

In yet another embodiment, wherein determining the final path according to the grid locations where the entry point and the exit point are located in the path cache includes: in response to the entry point and the exit point in the path cache being located in the same grid , performing a path planning operation on the entry point and the exit point to generate a final path; or in response to the entry point and the exit point in the path cache being in different grids, determining the final path based on finding a shared path.

In yet another embodiment, determining the final path based on searching for the shared path includes: searching for a shared path based on grid locations where the entry point and exit point are located; determining a start point and an end point of the shared path; The starting point and the ending point of , respectively perform a path planning operation with the entry point and the exit point to generate a splicing path; and splicing the shared path and the splicing path to generate the final path.

In yet another embodiment, wherein the path planning operation is performed by: determining slopes at adjacent waypoints in the path; connecting pathpoints with different slopes and the furthest distances to achieve path planning; and/ Or add new waypoints in a path segment containing multiple grid areas; connect the new waypoints to realize path planning.

In yet another embodiment, wherein determining the final path based on the path planning request and the path cache in the edge server comprises: determining a target USV within the coverage area of the edge server based on the path planning request the starting point and ending point of the target unmanned surface vehicle; query whether there is a path including the starting point and ending point of the target unmanned surface craft in the path cache in the edge server; and determine the final path according to the query result.

In yet another embodiment, the method further includes updating the final path into the path cache by performing the following operations: based on the query frequency of the path cache, the calculation cost of planning the final path, and the location of the final path in the path cache. The occupied space in the path cache determines the unit space revenue index of the final path; and the final path is updated into the path cache according to the storage capacity of the path cache and the unit space revenue index of the final path.

In another aspect, the present disclosure also provides a system for path planning for an unmanned surface vehicle, comprising: a plurality of edge servers for executing the method according to any one of claims 1-8; and a cloud The server is configured to: receive a path planning request sent by the multiple edge servers; generate a preliminary path plan based on the path planning request; and return the preliminary path plan to the corresponding edge server.

In yet another aspect, the present disclosure also provides a computer-readable storage medium having computer-readable instructions for path planning for an unmanned surface craft stored thereon, the computer-readable instructions being executed by one or more processors , to implement the above-mentioned multiple embodiments.

Through the solution of the present disclosure, the edge server and the cloud server are coordinated to perform path planning for the unmanned surface vehicle, which avoids the path planning calculation caused by the increase of the resolution of the map and the distance between the starting point and the target point of the unmanned surface vehicle. The time consumption increases exponentially, thereby improving the efficiency of path planning. Further, the solution of the present disclosure assists the path planning algorithm in determining the path by assisting in determining the final path through the path cache, thereby improving the response speed of the path request. In addition, the present disclosure also updates the path cache by calculating the unit space yield index of the final path, so as to improve the utilization rate and query rate of the path cache.

Description of drawings

The above and other objects, features and advantages of exemplary embodiments of the present disclosure will become readily understood by reading the following detailed description with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present disclosure are shown by way of example and not limitation, and like or corresponding reference numerals refer to like or corresponding parts wherein:

1 is an exemplary flowchart illustrating a method for path planning for an unmanned surface vehicle according to an embodiment of the present disclosure;

FIG. 2 is an exemplary schematic diagram illustrating a preliminary planned path according to an embodiment of the present disclosure;

3 is an exemplary schematic diagram illustrating a cache path according to an embodiment of the present disclosure;

4 is an exemplary schematic diagram illustrating path cache storage according to an embodiment of the present disclosure;

5 is an exemplary schematic diagram illustrating a path planning operation according to an embodiment of the present disclosure;

6 is a block diagram illustrating an exemplary flow chart for path planning for an unmanned surface vehicle according to an embodiment of the present disclosure;

7 is an exemplary flowchart illustrating a query path cache determining a final path according to an embodiment of the present disclosure;

FIG. 8 is an exemplary flowchart illustrating a cache update according to an embodiment of the present disclosure; and

9 is an exemplary schematic diagram illustrating a system for path planning for an unmanned surface craft according to an embodiment of the present disclosure.

Detailed ways

The technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the embodiments described in this specification are only some of the embodiments provided by the present disclosure for the purpose of facilitating the clear understanding of the solution and complying with legal requirements, and not all embodiments of the present disclosure can be implemented. Based on the embodiments disclosed in this specification, all other embodiments obtained by those skilled in the art without creative efforts shall fall within the protection scope of the present disclosure.

FIG. 1 is an exemplary flow diagram illustrating a method 100 for path planning for an unmanned surface vehicle according to an embodiment of the present disclosure. In one implementation scenario, the method 100 herein may be performed by an edge server.

As shown in FIG. 1, at step S102, a path planning request sent from the target unmanned surface vessel is received. Next, at step S104, the final path of the target unmanned surface vehicle is determined within or outside the coverage area of the edge server according to the path planning request, so that the target unmanned surface vehicle moves according to the final path. In an implementation scenario, the aforementioned edge servers may include one or more, and the coverage area of the edge servers may be a map area that is artificially divided according to the signal coverage of each edge server (for example, according to the coverage of the edge servers in FIG. 9 ) The range divides the total map area M into map area M1 and map area M2). It can be understood that the aforementioned path planning request is within the coverage area of the edge server means that the starting point and the end point of the path planning request are within the map area covered by the edge server. The aforementioned path planning request is outside the coverage area of the edge server means that the start point of the path planning request is within the map area covered by the edge server, and the end point is outside the map area covered by the edge server.

Specifically, the above-mentioned step S104 may include in step S104-1 (shown by a dashed box) and step S104-2 (shown by a dashed box) respectively according to the path planning request outside the coverage area of the edge server and the path planning request at Two scenarios within the coverage area of the edge server determine the final path of the target unmanned surface vehicle.

In one embodiment, when the path planning request is outside the coverage area of the edge server, the above step S104-1 further includes: in step S104-11, sending a path planning request to the cloud server to obtain the cloud server-based path planning request Returns the preliminary path plan. In some embodiments, the preliminary path planning may include entry points and exit points passing through the coverage area of the corresponding edge server obtained by the cloud server based on the path planning request, and the preliminary path planning will be described in detail later with reference to FIG. 2 . Based on the received preliminary path plan, in step S104-12, a final path is determined based on the preliminary path plan and the path cache in the edge server. As mentioned above, the preliminary path planning includes the entry points and exit points passing through the coverage area of the corresponding edge server, so that the path cache can be queried whether there is a path including the entry points and exit points, and the final path can be determined based on the query results. .

In one implementation scenario, when it is queried that there is a path including the entry point and exit point passing through the coverage area of the corresponding edge server in the path cache, the path passing through the aforementioned entry point and exit point in the path cache is used as the final path. In another implementation scenario, when there is no path including the entry point and exit point passing through the coverage area of the corresponding edge server in the query path cache, the final path is determined according to the grid information of the entry point and the exit point in the path cache. In some embodiments, the aforementioned entry point and exit point may be in the same grid or in different grids, which will be described in detail below for these two scenarios respectively.

In one scenario, a final path is determined based on finding a shared path in response to entry and exit points in the path cache being in different meshes. Specifically, first, the shared path can be searched based on the grid positions of the entry point and the exit point, and the start point and end point of the shared path can be determined. Next, a path planning operation is performed on the start point and the end point of the shared path, and the entry point and the exit point, respectively, to generate a spliced path. Finally, the shared path and the spliced path are spliced to generate the final path. In one embodiment, the aforementioned determination of the final path based on finding the shared path can be represented by the following mathematical expression:

P _s,t = {P _s,a ,P _a,b +P _b,t } (1)

Among them, P _s,t represents the final path, P _a,b represents the shared path with a and b as the starting point and end point, P _s,a , P _b,t represents the spliced path obtained by performing the path planning operation on the edge server . Further, the final path length can be expressed as the following formula:

D _s,t =D _s,a +D _a,b +D _b,t (2)

Among them, D _s,t represents the final path length, D _a,b represents the shared path length, D _s,a , D _b,t represent the splicing path length. It can be seen from formula (2) that the length error of the final path is inversely proportional to the length of the shared path, that is, when the length of the shared path is much greater than the sum of the lengths of the spliced paths, the error can be ignored. The aforementioned searching for a shared path to determine the final path and the aforementioned path planning operations will be described in detail later with reference to FIGS. 4-5 .

In another scenario, in response to the entry point and exit point in the path cache being in the same mesh, the path planning operation is performed on the entry point and the exit point to generate the final path, that is, the path planning operation is performed directly through the edge server in order to plan a path A path with the entry and exit points as the start and end points is the final path.

In one embodiment, when the path planning request is within the coverage area of the edge server, the above step S104-2 further includes step S104-21: determining the final path based on the path planning request and the path cache in the edge server, so that the target is unattended The surface craft follows the final path. Specifically, the starting point and the ending point of the target unmanned surface vehicle within the coverage area of the edge server can be determined first based on the path planning request. Similar to the preliminary path planning returned by the cloud server obtained above, the path cache can then be queried to see if there is a path including the starting point and the ending point of the target unmanned surface vessel, and then the final path can be determined according to the query result. For details, refer to the content described in the foregoing step S104-12.

It can be seen from the above description that the solution of the present disclosure performs path planning for the target unmanned surface vessel through collaboration between the edge server and the cloud server. It can be understood that the cloud server performs preliminary path planning for the target unmanned surface vehicle through a low-resolution map. Further, the cloud server returns the preliminary path planning to the corresponding edge server (equivalent to each edge server receiving a path request sent from the cloud server), and then uses the high-resolution map to execute the path planning through the edge server, thereby avoiding Therefore, due to the increase of the map resolution and the distance between the starting point and the target point of the unmanned surface vehicle, the path planning calculation time increases exponentially, and the path planning efficiency is improved. In addition, the present disclosure also determines the path by querying the path cache, so that the response speed of the path request is greatly improved.

FIG. 2 is an exemplary schematic diagram illustrating a preliminary planned path according to an embodiment of the present disclosure. As shown in FIG. 2, areas A, B, C, and D divided by four rectangular boxes are shown in sequence, and the areas A, B, C, and D may be coverage areas corresponding to four edge servers. Assume that the starting point of the path planning request sent by the target unmanned surface craft is in area A, and denoted as point v _s ; the end point is in area D, and denoted as point v _t . That is, the path planning request of the target unmanned surface vehicle is outside the coverage area of the edge server. In this scenario, according to the foregoing description, the aforementioned path planning request may be sent to the cloud server via the edge server. Next, the cloud server uses the low-resolution map to perform preliminary path planning for the target unmanned surface vessel to obtain a path P _s,t , which passes through areas A, C, and D in sequence, that is, passes through points v _s , v ₁ , v ₂ and the path of v _t . After preliminary planning by the cloud server, point v ₁ is the exit point and entry point of area A and area C, and point v ₂ is the exit point and entry point of area C and area D. In an exemplary scenario, the cloud server returns point v _s and exit point v ₁ to the edge server corresponding to area A. Similarly, return entry point v1 and exit point _v2 to the edge server corresponding to region C and _return entry point _v2 and point _vt to the edge server corresponding to region C. Further, each edge server determines the final path according to the received entry point, exit point and path cache.

According to the entry point and exit point returned by the cloud server to the corresponding edge server, the edge server first queries whether there is a path including the entry point and exit point in the path cache, and then determines the final path based on the query result. As mentioned above, when there is a path including the entry point and the exit point in the path cache, the path passing through the entry point and the exit point in the path cache is used as the final path. When there is no path including the entry point and the exit point in the query path cache, the final path is determined according to the grid information of the entry point and the exit point in the path cache. The query path cache will be described in detail below with reference to FIG. 3 .

FIG. 3 is an exemplary schematic diagram illustrating a cache path according to an embodiment of the present disclosure. As shown in FIG. 3 , it is assumed that passing through points v ₁ , v ₂ , v ₃ , v ₄ and v ₅ in sequence is a cache path, and passing through points v ₇ , v ₆ , v ₄ and v ₉ in sequence is another cache path. In an exemplary scenario, assuming that point _v7 and point _v9 respectively represent the entry point and exit point returned to the area covered by the edge server via the cloud server, it is determined by querying the path cache that there is a path including passing through point _v7 and point _v9 . path, the path passing through point v ₇ and point v ₉ can be used as the final path. Also, the cache paths passing through v ₇ , v ₆ , v ₄ and v ₉ are taken as the final path.

The point v8 is further shown in the figure, assuming that the point _v8 and the point _{v9 respectively} represent the entry point and the exit point returned to the area covered by the edge server via the cloud server, it is determined by querying the path cache that there is no point _v8 and For the path of point _v9 , the final path needs to be determined according to the grid information where point _v8 and point _v9 are located. As can be seen from the foregoing, when the entry point and the exit point are in the same grid, the path planning operation is performed on the entry point and the exit point to generate the final path. In an exemplary scenario, assuming that point _v8 and point _v7 are the entry point and exit point of the area covered by the edge server, _respectively , and point _v8 and point _v7 are in the same grid, then the Point 7 performs path planning to generate the final path. That is, a path from point v8 to point ₇ is generated (shown by the dotted line in the figure). When the entry and exit points in the path cache (eg, point _v8 and point _v9 ) are in different grids, the final path is determined based on finding the shared path.

In determining the final path based on finding the shared path, the shared path can be firstly searched based on the grid positions where the entry point and the exit point are located. Next, a path planning operation is performed on the start point and the end point of the shared path, and the entry point and the exit point, respectively, to generate a spliced path. Finally, the shared path and the spliced path are spliced to generate the final path. In one embodiment, in order to facilitate fast search of shared paths and improve search efficiency, the present disclosure also improves the storage method of the path cache, that is, by storing the grid positions of the path nodes, and updating the grid based on the stored grid positions of the path nodes Grid reverse list. In an implementation scenario, the foregoing mesh inversion list may store all path information of mesh boundary points. The storage mode of the path cache will be described in detail below with reference to FIG. 4 .

FIG. 4 is an exemplary schematic diagram illustrating path cache storage according to an embodiment of the present disclosure. Taking the above Figure 3 as an example again, suppose there are two paths P _1,5 and P _7,9 , and the path P _1,5 passes through the points v ₁ , v ₂ , v ₃ , v ₄ and v ₅ , where the point v ₁ are in grid _G1 , _v2 and _v3 are in grid _G2 , _v4 is in grid _G5 , and _v5 is in grid _G8 . The aforementioned path P _7,9 passes through points v ₇ , v ₆ , v ₄ and v ₉ , where points v ₇ and v ₆ are in grid G ₄ and v ₉ is in grid G ₃ . In an implementation scenario, each path and its path nodes can be stored as shown in the left figure in Figure 4, that is, the path node information corresponding to the paths P _1,5 are G ₁ (v ₁ ), G ₂ (v ₂ , v ₃ ), G ₅ (v ₄ ), and G ₈ (v ₅ ). Similarly, the path node information corresponding to the paths P ₇ and 9 are G ₄ (v ₇ , v ₆ ), G ₅ (v ₄ ) and G ₃ (v ₉ ). Based on the aforementioned stored path information, the mesh reverse list shown on the right side of FIG. 4 may be further updated, and the mesh reverse list may include mesh numbers and corresponding paths. For example, grid G ₁ , grid G ₂ , and grid G ₅ all correspond to path P _1,5 , grid G ₃ , grid G ₄ , and grid G ₅ .

Through the storage method of the above path cache, the shared path can be quickly located. Still taking the above FIG. 3 as an example, it is assumed that the final path takes the point _v8 and the point _v9 as the entry point and the exit point respectively, and the path nodes are stored in the above-mentioned manner. Since point _v8 is within grid _{G4, which corresponds to path P7,9 ,} path _P7,9 can be _used as a shared path. Further, the starting point and the ending point are determined according to the searched shared path. In one embodiment, an optimal path can be determined by calculating the distance from the entry point to the exit point to determine the starting point and the ending point of the shared path. For example, the distance from point v ₈ to point v ₇ and point v ₆ can be calculated. If point v ₈ is closer to point v ₇ , point v ₇ can be used as the starting point of the shared path. Similarly, the end point of the shared path can be determined to be _v9 . Next, a path planning operation is performed on the start point and the end point of the shared path, and the entry point and the exit point, respectively, to generate a spliced path. For example, taking the entry point v ₈ and the starting point v ₇ of the shared path as an example, the path planning operation is performed on the point v ₈ and the point v ₇ to generate a spliced path (eg, as shown by the dotted line in FIG. 3 ). Finally, the shared paths P ₇ , 9 and the spliced paths are spliced to form the final path.

的整数倍，并且通常是向相邻的8个栅格前进至下一步。此外，传统A*算法获得的路径转折点较多、路径不够平滑，且储存的路径节点较多，从而不利于存储路径以及查找共享路径。在有障碍物的地图中，传统A*算法规划出的路径通常由斜向、水平、垂直三个方向的线段交替出现，而三角形任意两边之和大于第三边。有鉴于此，本公开的方案是通过尽可能的采用一条边表示原来两条边的路径，同时只保留首尾两个路径节点。在一个实现场景中，可以通过确定路径中相邻路径点处的斜率，从而将斜率不同并且距离最远的路径点的进行连接，以实现采用一条边表示原来两条边的路径。例如假设有一条路径P_a,b＝{c₁,c₂,…,c_m}，其相邻两个路径节点的斜率定义为α(c_i-1,c_i)。当α(c_i-1,c_i)≠α(c_i,c_i+1)时，可以判断对应边E(c_i-1,c_i)的斜率与边E(c_i,c_i+1)的斜率不同，则路径节点c_i为两条边的转折点，并且将两条边上分布较远的转折点进行连接且不穿越障碍物。下面将结合图5详细描述经由改进A*算法实现上述路径规划操作。In one embodiment, the present disclosure also improves the A* algorithm to implement the above path planning operation. It can be understood that the operation angle of the traditional A* algorithm is limited to

an integer multiple of , and usually advances to the next 8 grids. In addition, the path obtained by the traditional A* algorithm has many turning points, the path is not smooth enough, and there are many path nodes stored, which is not conducive to storing paths and finding shared paths. In a map with obstacles, the path planned by the traditional A* algorithm usually consists of line segments in the diagonal, horizontal, and vertical directions alternately, and the sum of any two sides of the triangle is greater than the third side. In view of this, the solution of the present disclosure is to use one edge as much as possible to represent the path of the original two edges, while retaining only the first and last two path nodes. In an implementation scenario, the slopes at adjacent waypoints in the path can be determined, so as to connect the waypoints with different slopes and the farthest distance, so as to realize a path using one edge to represent the original two edges. For example, suppose there is a path P _a,b = _{ c ₁ ,c ₂ ,...,cm }, and the slopes of its two adjacent path nodes are defined as α( _ci-1 , _ci ). When α( _ci-1 ,ci) _≠ α( _ci , _ci+1 ), the slope of the corresponding side E( _ci-1 , _ci ) can be judged and the side E( _ci , _{ci+ 1} ) with different slopes, then the path node c _i is the turning point of the two edges, and the far-distributed turning points on the two edges are connected without passing through obstacles. The implementation of the above path planning operation via the improved A* algorithm will be described in detail below with reference to FIG. 5 .

FIG. 5 is an exemplary schematic diagram illustrating a path planning operation according to an embodiment of the present disclosure. As shown on the left side of Fig. 5, the path P _a,b obtained based on the path planning performed by the traditional A* algorithm, the path P _a,b passes through multiple path nodes and bypasses obstacles (such as the black square in the figure). ). Taking point c1 to point c6 as an example, the slope at point c1 is the same as that at point c2, the slope at point c3 to point c6 is the same but different from the slopes of points c1 and c2, and point c1 is closer to point c6. Far. It is thus possible to connect point c1 with point c6. Similarly, point c7 can be connected with point c8, point c9 and point c10. Based on the foregoing operations, the path shown on the right side of the figure can be obtained. In one implementation scenario, when a path segment traverses multiple grids, such as the path segment from point c11 to point c12 shown in the figure, spans three grids. In this scenario, a path planning operation can be implemented by adding new path nodes (for example, point D on the path shown on the right side of the figure) and connecting the new path points. The path nodes, turning points and the total length of the path after performing the path planning operation using the traditional A* algorithm and the improved A* algorithm are shown in Table 1 below.

Table 1 Comparison of path planning operations performed by traditional A* algorithm and improved A* algorithm

algorithm number of nodes number of turns total length of the path Traditional A* 25 9 28.97 Improve A* 11 8 23.77

It can be seen from Table 1 above that compared with the traditional A* algorithm, the use of the improved A* algorithm to perform path planning operations can reduce unnecessary path nodes and turning points, and shorten the total length of the path, thereby reducing the storage space of the path cache. And improve the sharing ability of the path.

FIG. 6 is a block diagram illustrating an exemplary flowchart for path planning for an unmanned surface vehicle according to an embodiment of the present disclosure. It should be understood that FIG. 6 is a specific embodiment of the method 100 in the foregoing FIG. 1 , so the above descriptions with respect to FIG. 1 are also applicable to FIG. 6 .

As shown in FIG. 6, at step S601, a path request is sent by the unmanned surface vessel to the edge server. Based on the aforementioned path request, at step S602, the edge server determines whether the path request is within its coverage area according to the received path request. When the route request is within the coverage area of the current edge server, the flow proceeds to step S603. At this step, the final path is determined by looking up the path cache. When the path request is outside the coverage area of the current edge server, the aforementioned path request is sent to the cloud server via the edge server, and at step S604, preliminary path planning is performed by the cloud server to obtain the entry point and exit point corresponding to the edge server . According to the obtained entry point and exit point, the flow jumps to step S603. That is, the final path is determined by looking up the path cache. Further, at step S605, it is determined whether the path is hit in the path cache, for example, whether the path cache contains a path passing through the entry point and the exit point. When the path is hit, at step S606, the edge server returns the hit path as the final path to the target unmanned surface vehicle (ie, responds to the aforementioned path planning request). At step S607, the target unmanned surface vehicle executes the path according to the final path returned by the edge server (ie moves according to the final path). As further shown in the figure, when a path is missed, at step S608, a path planning operation is performed by finding a shared path and using the modified A* algorithm to determine the final path. Regarding the foregoing search for a shared path and path planning operations, reference may be made to the content described in the foregoing FIG. 3 to FIG. 5 , and details are not repeated here. Finally, at step S609, the final path is updated into the path cache.

FIG. 7 is an exemplary flowchart illustrating a query path cache determining a final path according to an embodiment of the present disclosure. As shown in FIG. 7, at step S701, the path cache is queried according to the entry point (or start point) and the exit point (or end point) of the corresponding edge server. Next, at step S702, it is determined whether the path cache is empty. That is, by querying whether the path cache contains paths passing through the entry point and exit point of the corresponding edge server. When the path cache does not contain the path passing through the entry point and the exit point of the corresponding edge server (that is, the path cache is empty), at step S703, the edge server uses the improved A* algorithm to perform path planning for the entry point and the exit point operation to get the final path. Further, at step S704, the final path result is returned.

When the path cache contains paths passing through the entry point and exit point of the corresponding edge server (that is, the path cache is not empty), at step S705, it is determined whether the entry point and the exit point of the corresponding edge server are in the same grid. When the entry point and the exit point corresponding to the edge server are in the same grid, the process jumps to S703, that is, the edge server uses the improved A* algorithm to perform a path planning operation on the entry point and the exit point, and at step S704, return to the final path result. When the entry point and the exit point of the corresponding edge server are in different grids, at step S706, the candidate path list PL is filtered out from the path cache, so as to find the shared path.

Based on the above-screened candidate path list PL, at step S707, it is determined whether the candidate path list PL is empty. Similarly, when the candidate path list PL is an empty list, the flow jumps to S703, and returns to the final path result via step S704. When the candidate path list PL is not an empty list, at step S708, the optimal shared path is searched. Next, at S709, a splicing path is determined based on the entry point and exit point of the corresponding edge server and the start point and end point of the shared path. The shared path and the spliced path are spliced to generate a final path, and in step S704, the final path result is returned.

In one embodiment, the present disclosure may further include updating the final path to the path cache (eg, step S609 in FIG. 6 above). Specifically, the unit space revenue index of the final path may be determined based on the query frequency of the path cache, the calculation cost of planning the final path, and the space occupied by the final path in the path cache. Next, the final route is updated into the route cache according to the storage capacity of the route cache and the unit space revenue index of the final route.

In an implementation scenario, it is assumed that the query frequency of the path cache is denoted as F _s,t , and the calculation cost of planning the final path is denoted as C _s,t . It can be understood that this computational cost grows exponentially as the distance between the source point and the target point increases. Suppose the space occupied by the final path in the path cache is denoted as S _s,t , and S _s,t can be expressed as the following formula:

S _s,t =|P _s,t +2·|G _s,t || (3)

Among them, S _{s, t} represents the total occupied space of the cache path P _{s, t} , P _{s, t} represents the path information of all cache paths in the path cache, and G _{s, t} represents the grid information that all cache paths pass through.

则

可以通过如下式子表示：According to the query frequency F _s,t obtained above, the calculation cost C _s,t and the occupied space S _s,t , the unit space benefit index of the final path can be determined. Suppose the unit space revenue index is denoted as

but

It can be expressed by the following formula:

In an implementation scenario, since the goal of cache construction and update is to achieve a cache structure that maximizes the total revenue, the storage space of the edge server is limited. Therefore, the problem can be expressed as the following formula:

表示单位空间收益指标，x_i表示具体路径，m表示当前缓存的个数，

表示缓存总容量。上述公式(5)可以理解成一个NP完整的0-1背包问题，在本公开实施例中即如何将路径在有限存储空间进行缓存，以实现总收益最大化。例如在缓存空间足够时，直接将由边缘服务器执行的路径规划操作获得的全新路径(即不是根据查找共享路径确定的路径)插入缓存空间，并对缓存空间中的所有路径进行收益排序。在缓存空间已满时，则对全新路径和缓存路径按照收益进行排序，接着使用贪心算法选择需要缓存的路径，进而返回新的缓存结果。下面将结合图8详细描述前述缓存更新过程。in,

Represents the unit space revenue index, x _i represents the specific path, m represents the current number of caches,

Indicates the total cache capacity. The above formula (5) can be understood as an NP complete 0-1 knapsack problem, in the embodiment of the present disclosure, how to cache the path in a limited storage space to maximize the total revenue. For example, when the cache space is sufficient, a brand-new path obtained by the path planning operation performed by the edge server (that is, the path not determined by finding the shared path) is directly inserted into the cache space, and all paths in the cache space are sorted by revenue. When the cache space is full, the new paths and the cached paths are sorted according to the revenue, and then the greedy algorithm is used to select the paths that need to be cached, and then the new cached results are returned. The foregoing cache update process will be described in detail below with reference to FIG. 8 .

基于获得的单位空间收益指标，在步骤S804处，根据单位空间收益指标

对缓存空间ψ中的所有路径进行排序。最后，在步骤S805处，返回缓存空间ψ的缓存结果。FIG. 8 is an exemplary flowchart illustrating a cache update according to an embodiment of the present disclosure. As shown in FIG. 8, at step S801, a path P _s,t obtained by performing a path planning operation via the edge server is received. Next, at step S802, it is determined whether the cache space (ie, the path cache) ψ is full. When the cache space ψ is sufficient, at step S803, the path P _s,t can be directly inserted into the cache space ψ, and the unit space revenue index of the path P _s,t can be calculated

Based on the obtained unit space income index, at step S804, according to the unit space income index

Sort all paths in cache space ψ. Finally, at step S805, the cached result of the cache space ψ is returned.

和

进一步地，在步骤S807处，判断全新的路径P_s,t的单位空间收益指标

是否大于已有缓存路径的单位空间收益指标

当

大于

时，在步骤S808处，利用全新的路径P_s,t替换已有缓存路径，接着流程跳转至步骤S804至步骤S805。与之相反地，当

小于

时，流程跳转至步骤S805，以返回缓存空间ψ的缓存结果。When the cache space ψ is full, at step S806, calculate the path P _s,t and the unit space revenue index of the existing cache path respectively

and

Further, at step S807, determine the unit space income index of the new path P _s,t

Whether it is greater than the unit space revenue index of the existing cache path

when

more than the

, at step S808, replace the existing cache path with a brand new path P _s,t , and then the flow jumps to step S804 to step S805. On the contrary, when

less than

, the flow jumps to step S805 to return the cached result of the cache space ψ.

It can be seen from the above description that the solution of the present disclosure reduces the number of accesses to the cloud server by coordinating path planning between the edge server and the cloud server, not only reduces the network delay, but also improves the path planning efficiency. Further, the present disclosure divides the complete map into multiple sub-maps and saves them on the edge server, so that the path planning delay can be reduced. Further, the present disclosure also utilizes the path auxiliary cache to determine the final path, which reduces the path planning delay, thereby improving the response speed of the path request. In addition, the embodiments of the present disclosure also use the improved A* algorithm to perform the path planning operation, so as to reduce the path storage space and obtain a smoother path. In some embodiments, the present disclosure also updates the cache based on the unit space revenue index of the final path, so as to ensure a cache structure that maximizes the total revenue in limited storage space.

9 is an exemplary schematic diagram illustrating a system 900 for path planning of an unmanned surface craft according to an embodiment of the present disclosure. As shown in FIG. 9 , the system 900 may include a plurality of edge servers 901 and cloud servers 902 . In one implementation scenario, each edge server can communicate with the cloud server through a wired connection and through a mobile data network (eg, a "4G" or "5G" wireless cellular network). In one embodiment, the aforementioned cloud server 902 may be configured to receive path planning requests sent by multiple edge servers, and then generate a preliminary path plan based on the path planning requests. Finally, the preliminary path planning is returned to the corresponding edge server. In another embodiment, each edge server 901 performs the method described above in FIG. 1 based on the preliminary path planning returned by the cloud server, generates a final path, and returns the final path to the unmanned surface vehicle 903 so that no Surface craft 903 moves according to the final path. In some embodiments, the system 900 of the present disclosure may also include a Road Side Unit (“RSU”) 904 . The RSU 904 can be used for communication between the edge server and the unmanned surface vehicle.

From the above description in conjunction with the accompanying drawings, those skilled in the art can also understand that the embodiments of the present disclosure can also be implemented by software programs. The present disclosure thus also provides a computer program product. The computer program product can be used to implement the method for path planning for an unmanned surface craft described in the present disclosure in conjunction with FIGS. 1-8.

It should be noted that although the operations of the disclosed methods are depicted in the figures in a particular order, this does not require or imply that the operations must be performed in the particular order, or that all illustrated operations must be performed to achieve desirable results. Rather, the steps depicted in the flowcharts may change the order of execution. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined to be performed as one step, and/or one step may be decomposed into multiple steps to be performed.

It should be understood that when the terms "first", "second", "third" and "fourth" are used in the claims of the present disclosure, the description and the drawings, they are only used to distinguish different objects, and Not intended to describe a specific order. The terms "comprising" and "comprising" as used in the specification and claims of the present disclosure indicate the presence of the described feature, integer, step, operation, element and/or component, but do not exclude one or more other features, integers , step, operation, element, component and/or the presence or addition of a collection thereof.

It should also be understood that the terminology used in the present disclosure is for the purpose of describing particular embodiments only and is not intended to limit the present disclosure. As used in this disclosure and the claims, the singular forms "a," "an," and "the" are intended to include the plural unless the context clearly dictates otherwise. It should further be understood that, as used in this disclosure and the claims, the term "and/or" refers to and including any and all possible combinations of one or more of the associated listed items.

Although the embodiments of the present disclosure are as described above, the contents described are only examples adopted to facilitate understanding of the present disclosure, and are not intended to limit the scope and application scenarios of the present disclosure. Any person skilled in the technical field described in this disclosure, without departing from the spirit and scope disclosed in this disclosure, can make any modifications and changes in the form and details of implementation, but the scope of patent protection of this disclosure , still subject to the scope defined by the appended claims.

Claims (10)

1. A method for path planning for an unmanned surface craft, the method being performed by an edge server, and comprising: Receive the path planning request sent from the target unmanned surface vehicle; determining the final path of the target unmanned surface vehicle within or outside the coverage area of the edge server according to the path planning request, Wherein, when the path planning request is outside the coverage area of the edge server, the method includes: sending the path planning request to a cloud server to obtain a preliminary path plan returned by the cloud server based on the path planning request; and Based on the preliminary path planning and the path cache in the edge server determine the most final path; Wherein, when the path planning request is within the coverage area of the edge server, the method includes: A final path is determined based on the path planning request and the path cache in the edge server, so that the target unmanned surface vehicle moves according to the final path. 2. The method of claim 1, wherein the preliminary path planning includes entry points and exit points through a coverage area of a corresponding edge server, wherein the final determination is based on the preliminary path planning and path caching in the edge server. Paths include: querying the path cache for the existence of a path containing the entry point and exit point; and The final path is determined based on the query results. 3. The method of claim 2, wherein determining the final path based on the query results comprises: When querying the path cache for a path including the entry point and the exit point, take the path passing through the entry point and the exit point in the path cache as the final path; or When it is queried that there is no path including the entry point and the exit point in the path cache, the final path is determined according to the grid information in the path cache where the entry point and the exit point are located. 4. The method according to claim 3, wherein determining the final path according to the grid positions where the entry point and the exit point are located in the path cache comprises: In response to the entry point and exit point in the path cache being within the same grid, performing a path planning operation on the entry point and exit point to generate a final path; or A final path is determined based on finding a shared path in response to the entry and exit points in the path cache being in different meshes. 5. The method of claim 4, wherein determining the final path based on finding the shared path comprises: Find a shared path based on the grid location where the entry point and the exit point are located; determining the starting point and the ending point of the shared path; performing a path planning operation on the start point and the end point of the shared path and the entry point and the exit point, respectively, to generate a spliced path; and The shared path is spliced with the spliced path to generate the final path. 6. The method of claim 5, wherein the path planning operation is performed by: Determine the slope at adjacent waypoints in the path; Connecting path points with different slopes and the furthest distances for path planning; and/or Add new waypoints to path segments containing multiple mesh regions; The new waypoints are connected to realize route planning. 7. The method of claim 1, wherein determining a final path based on the path planning request and a path cache in the edge server comprises: determining, based on the path planning request, the starting point and the ending point of the target unmanned surface vehicle within the coverage area of the edge server; querying the path cache in the edge server whether there is a path containing the origin and destination of the target unmanned surface vehicle; and Determine the final path based on the query result. 8. The method according to any one of claims 1-7, further comprising performing the following operations to update the final path into the path cache: Determine the unit space revenue index of the final path based on the query frequency of the path cache, the calculation cost of planning the final path, and the space occupied by the final path in the path cache; The final route is updated into the route cache according to the storage capacity of the route cache and the unit space revenue index of the final route. 9. A system for path planning for an unmanned surface craft, comprising: a plurality of edge servers for performing the method of any one of claims 1-8; and Cloud servers, which are used for: receiving a path planning request sent by the multiple edge servers; generating a preliminary path plan based on the path planning request; and The preliminary path planning is returned to the corresponding edge server. 10. A computer readable storage medium comprising computer program instructions for path planning for an unmanned surface craft, which when executed by one or more processors, cause the implementation according to claim 1 . The method described in -8.

Images