A dynamic path planning method for a cluster of unmanned aerial vehicles for hull corrosion detection

Abstract

The application relates to a kind of unmanned aerial vehicle cluster dynamic path planning method for ship body corrosion detection, comprising the following steps: utilizing high-precision sensor scanning ship body surface, generating three-dimensional model, dividing corrosion area based on corrosion prediction model;Based on the corrosion area divided, the task area distribution is dynamically adjusted by DARP algorithm, and the task area is allocated to each unmanned aerial vehicle, and the task area priority is calculated;Based on the task area allocated, the path of unmanned aerial vehicle is planned by improved A* algorithm, combined with task area priority, and the path is smoothed and optimized by bidirectional Floyd algorithm;In the process of ship body corrosion detection, real-time update ship body corrosion information fed back by unmanned aerial vehicle, dynamically adjust task area priority, modify the path of unmanned aerial vehicle.Compared with prior art, the application has the advantages of improving detection efficiency, avoiding resource waste and the like.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned aerial vehicle cluster path planning, and particularly relates to a dynamic path planning method for unmanned aerial vehicle cluster for ship hull corrosion detection. BACKGROUND

[0002] Under the background of economic globalization in today's era, shipping industry has become an important foundation for economic and social development, and corrosion of ship hull is an inevitable problem in the shipping industry, and is also an important factor affecting the safety and service life of ships. Therefore, regular corrosion detection on the surface of the ship hull is particularly important to ensure the structural safety of the ship. However, the traditional ship hull corrosion detection method mostly relies on manual inspection or single automatic detection equipment, and these methods are not only low in efficiency, high in cost and complex in operation. Manual inspection not only requires a large amount of time and manpower, and in the harsh marine environment, the safety problem cannot be guaranteed. Even though the single robot detection technology has certain advantages, it still faces problems such as low path planning efficiency and uneven task allocation, especially in the face of complex ship surface and uneven corrosion area distribution, the traditional path planning algorithm often cannot respond to dynamic changes in real time, resulting in low resource utilization and low detection efficiency.

[0003] The existing unmanned aerial vehicle cluster path planning technology mainly relies on static regional division method, such as dividing the detection area into several sub-regions, and then assigning these regions to different unmanned aerial vehicles for independent detection. However, this method cannot respond to the dynamic changes of the corrosion area and make timely feedback, which may cause some corrosion areas to be missed or multiple unmanned aerial vehicles to detect the same area repeatedly, resulting in unnecessary waste of resources. At the same time, the traditional path planning method also lacks the adaptability to the complex ship curved surface and corrosion distribution characteristics, resulting in low efficiency of unmanned aerial vehicle flight path planning, and unable to effectively reduce the flight time and energy consumption.

[0004] CN116048119A discloses an unmanned aerial vehicle cruise monitoring system and method based on artificial intelligence. The system identifies the terrain feature information of the task area through satellite radar, divides the task area in two-dimensional plane coding, and further maps the cruise path in the two-dimensional plane coding, so that the cruise path can be accurately associated in the unified coding. The system combines the direction angle and the route distance in the static cruise path to deeply associate each cruise path, calculates the safety factor of the cruise path, and when problems occur in the dynamic cruise process, combines the changes of the dynamic direction angle and the route distance to quickly match the unmanned aerial vehicle on the most similar other route to compensate for the task operation, so as to unify the scheduling of the static cruise path and the dynamic task operation change, and multiple unmanned aerial vehicles and multiple routes, form a replacement decision, timely respond to unknown risks or unexpected situations, and make the unmanned aerial vehicle more flexible and changeable. Although the method realizes dynamic path planning, it is based on a static path, calls other unmanned aerial vehicles to compensate for the changes in the task, and is not a dynamic path planning for each unmanned aerial vehicle. Moreover, dynamic region division is not performed, and the unmanned aerial vehicle still performs dynamic optimization in the global range, which will lead to reduced detection efficiency and wasted unmanned aerial vehicle resources, and is not suitable for ship body corrosion detection.

[0005] Therefore, how to combine dynamic region allocation and autonomous path planning technology to realize efficient task allocation and path optimization in ship body corrosion detection, ensure that the corrosion area can be detected preferentially, and reduce redundant flight and energy waste, has become a key problem to be solved in the current technology. SUMMARY

[0006] The purpose of the present application is to provide a method for dynamic path planning of unmanned aerial vehicle cluster for ship body corrosion detection. By combining the DARP algorithm and the improved A* path planning algorithm, the method can realize efficient and flexible completion of corrosion detection tasks by the unmanned aerial vehicle cluster in a complex environment. The method can dynamically adjust the flight path according to the real-time feedback data, preferentially cover the corrosion hot area, and realize task cooperation between multiple unmanned aerial vehicles, so as to ensure efficient and accurate completion of comprehensive detection of the ship body surface.

[0007] The purpose of the present application can be achieved by the following technical solutions:

[0008] A method for dynamic path planning of unmanned aerial vehicle cluster for ship body corrosion detection, comprising the following steps:

[0009] Region division initialization: scan the ship body surface by using a high-precision sensor to generate a three-dimensional model, and divide the corrosion area based on a corrosion prediction model;

[0010] Region allocation: based on the divided corrosion regions, dynamically adjust the task region allocation through the DARP algorithm, and assign each UAV with a task region, calculate the task region priority;

[0011] Path planning and optimization: based on the allocated task region, use the improved A* algorithm to plan the UAV path, combined with the task region priority, and smooth the path through the bidirectional Floyd algorithm;

[0012] Dynamic priority adjustment and path modification: in the process of ship corrosion detection, real-time update the ship corrosion information feedback by the UAV, dynamically adjust the task region priority, and modify the UAV path.

[0013] The corrosion region division based on the corrosion prediction model specifically includes: preliminary analysis of the corrosion region through the corrosion prediction model, and generating a corrosion prediction matrix:

[0014]

[0015] Wherein, C i,j represents the corrosion prediction value of different regions of the ship.

[0016] The dynamic adjustment of task region allocation through the DARP algorithm specifically includes:

[0017] Divide n task regions R1, R2, …, R n and assign them to m UAVs, determine the task load function according to the corrosion area and density of the region, and assign the UAV based on the task load function, wherein the task load function is represented as follows:

[0018] L(R i )=αA(R i )+γρ(R i )

[0019] Wherein, L(R i ) represents the task load of task region R i , A(R i ) is the area of task region R i , ρ(R i ) is the corrosion distribution density of task region R i , and α and γ are the corresponding weight coefficients.

[0020] The dynamic adjustment of task region allocation further includes: dynamically adjusting the allocation of task regions according to the flight capability of the UAV, ensuring that the task load of each UAV is appropriate, and the flight capability of the UAV is determined by the following method:

[0021] Define the flight capability vector of the jth UAV as:

[0022] C j =(E j ,V j ,M j )

[0023] wherein E j is the percentage of the remaining power of the jth UAV, V j is the flight speed of the jth UAV, and M j is the maximum payload of the jth UAV;

[0024] After normalizing the flight capability vector, an integrated capability coefficient ψ j is obtained to represent the flight capability:

[0025] ψ j =α1E j +α2V j +α3M j

[0026] wherein α i is a weight,

[0027] The dynamic adjustment of the task area allocation satisfies the following constraint conditions:

[0028]

[0029] wherein m and n respectively represent the number of UAVs and the number of ship task areas, L(I j ) represents the load of the UAV I j , and L(R i ) represents the task load of the task area R i .

[0030] The cost function of the improved A* algorithm is:

[0031]

[0032] wherein f(n) is an integrated cost distance value, g(n) is the actual shortest cost distance from the starting node to the target node, h(n) is a heuristic function representing the cost distance from the current node to the target node, l is the distance from the current position to the target position, L is the distance from the starting point to the target point, λ is a coefficient for controlling the influence of the task area priority on path planning, and W(n) is a priority weight determined based on the task area priority.

[0033] The path optimization formula of the bidirectional Floyd algorithm is:

[0034] D[i][j]=min(D[i][j],D[i][k]+D[k][j])

[0035] wherein D[i][j] is the shortest path cost from node i to node j, and k is an intermediate node;

[0036] The shortest path cost matrix D is constantly updated to smooth the turns in the path.

[0037] The task area priority is represented as:

[0038] P i = αS corr (R i ) + βC path (R i )

[0039] wherein P i is the priority of the task area R i , S corr (R i ) is the corrosion severity of the area R i , expressed in terms of corrosion area or corrosion density, C path (R i ) is the path cost of the UAV from the current position to the task area, expressed in terms of flight distance or flight time, and α and β are adjustment coefficients.

[0040] In the dynamic priority adjustment and path modification, the path modification formula is as follows:

[0041] C new [i][j] = min(C old [i][j], C adjusted [i][j])

[0042] wherein C old [i][j] is the original path cost; C new [i][j] is the modified path cost; and C adjusted [i][j] is the path cost determined according to the new task area priority.

[0043] The method further comprises:

[0044] Multi-UAV coordination and conflict management: During the execution of the task by the UAV cluster, real-time detection of UAV path conflicts and flight time conflicts is performed, and when a conflict is detected, the path is adjusted to ensure that the paths do not intersect and minimize flight time and energy consumption.

[0045] Compared with the prior art, the present application has the following beneficial effects:

[0046] (1) The unmanned aerial vehicle cluster ship body corrosion detection method based on dynamic area allocation and path optimization provided by the application can realize omnidirectional detection of the ship body surface by the unmanned aerial vehicle cluster, so as to reduce the workload of manual detection and reduce the operation risk of manual detection, and improve the detection accuracy and efficiency. The method has the advantages of high efficiency, precision and automation, and is helpful for digital management and maintenance of the ship, and reduces the risk in the detection process.

[0047] (2) According to the corrosion condition of the ship body surface, the improved A* algorithm is adopted, and on this basis, the flight path is smoothed and optimized by combining the bidirectional Floyd algorithm, so that the unmanned aerial vehicle cluster can efficiently complete the detection of the ship body surface. And through the dynamic area allocation mechanism, the path overlap can be effectively reduced, so that the execution efficiency of the task is improved.

[0048] (3) The path planning and dynamic adjustment mechanism of the application can dynamically adjust the priority of the task area according to real-time data, so that the unmanned aerial vehicle can preferentially cover the area with serious corrosion degree. BRIEF DESCRIPTION OF DRAWINGS

[0049] Figure 1 is the flow chart of the method of the application;

[0050] Figure 2 is a ship task area division schematic diagram;

[0051] Figure 3 is the flow chart of the A* algorithm;

[0052] Figure 4 is the path optimization flow chart of the application. DETAILED DESCRIPTION

[0053] The application will be described in detail below in combination with the drawings and specific embodiments. The present embodiment is implemented on the basis of the technical scheme of the application, and gives a detailed implementation mode and specific operation process, but the protection scope of the application is not limited to the following examples.

[0054] The embodiment provides a UAV cluster dynamic path planning method for ship body corrosion detection, and aims to improve the efficiency and safety of ship body corrosion detection at sea. The method realizes flexible task allocation and efficient path planning of the UAV in a complex environment by fusing a DARP (Dynamic Area Region Partition) algorithm and an improved A* path planning algorithm. First, a high-precision sensor is used to scan and generate a three-dimensional model of the ship body, and a corrosion prediction model is constructed in combination with historical corrosion data and environmental factors or observations of an operator, so that preliminary corrosion hotspot analysis is performed. Then, the DARP algorithm is used to divide the ship body into multiple sub-regions (task regions), and each region is assigned to a UAV with different computing power according to the corrosion severity of the region, so that the task load of each UAV is balanced. On this basis, the bidirectional Floyd algorithm is applied to optimize the path smoothness to reduce flight bending and improve detection efficiency.

[0055] In addition, in view of the dynamic change characteristic of the corrosion region on the ship body surface, the method provides a mechanism for adjusting the priority of the task region in real time and modifying the flight path in time, so that the UAV can detect the corrosion severity flexibly and ensure that the high-priority region is detected in time. In order to ensure the cooperative work between the multiple UAVs and avoid conflicts, a path conflict detection and adjustment mechanism is also designed to ensure the effective cooperation of the UAV cluster.

[0056] Based on this, the method not only improves the efficiency of ship body corrosion detection, but also avoids the safety problems in manual detection, and has a positive effect on maritime communication.

[0057] Specifically, as shown in the figure, the method comprises the following steps: Figure 1

[0058] S1, region division initialization: a high-precision sensor is used to scan the surface of the ship body to generate a three-dimensional model, and corrosion regions are divided based on a corrosion prediction model.

[0059] First, a high-precision laser radar or sensor is used to scan the surface of the ship body to generate a three-dimensional model. Then, based on historical corrosion data, environmental influences and observations of a ground operator, a corrosion prediction model is constructed, and the corrosion regions are preliminarily analyzed by the corrosion prediction model to generate a corrosion prediction matrix:

[0060]

[0061] Wherein, C i,j represents the corrosion prediction value of different regions of the ship body, and -1 is a region known in advance. The matrix can evaluate the risk size of the different regions of the ship body affected by corrosion, and determine the potential high-risk corrosion region. ​

[0062] S2, region allocation: based on the divided corrosion region, the task region allocation is dynamically adjusted by the DARP algorithm, each UAV is allocated a task region, and the task region priority is calculated.

[0063] n task regions R1, R2, …, Rn are divided using the DARP algorithm (Dynamic Area Region Partition) n According to the corrosion distribution and the geometric shape of the ship surface, the size and shape of each region are dynamically adjusted to ensure the balanced task load of each UAV. Figure 2 The task region division result in an embodiment is shown. The divided task regions are allocated to m UAVs, i.e., different corrosion degree regions are divided to different UAVs, for example, the corrosion serious region is allocated to the UAV with strong computing power, the corrosion moderate region is allocated to the UAV with relatively strong computing power, and the corrosion low region is allocated to the UAV with low computing power. In this way, the balanced task load of each UAV is ensured, and the problem of some UAVs with heavy task load and low efficiency is avoided, which affects the whole task process. Further, the task load function is determined according to the corrosion area and density of the region, and the UAV is allocated based on the task load function, wherein the task load function is represented as follows:

[0064] L(R i )=αA(R i )+γρ(R i )

[0065] wherein L(R i ) represents the task load of the task region R i , A(R i ) is the area of the task region R i , ρ(R i ) is the corrosion distribution density of the task region R i , and α and γ are the corresponding weight coefficients. This formula enables the operator to weight the task by the area and the corrosion density. Through these calculations, the system can allocate appropriate task regions to each UAV to ensure the balance and efficiency of task execution.

[0066] In this embodiment, the task region allocation is also dynamically adjusted according to the flight capability of the UAV to ensure the appropriate task load of each UAV. To achieve adaptive allocation of task regions, the flight capability vector of the jth UAV is designed in this embodiment:

[0067] C j =(E j ,V j ,M j )

[0068] wherein Ej V is the percentage of remaining power, j M is the flight speed of the UAV, j M is the flight speed of the UAV,

[0069] ψ j = a1E j + a2V j + a3M j

[0070] wherein, According to the above task load function, the ability-load matching degree function is introduced:

[0071] Δ ij = |ψ j - L(R i )|

[0072] Then the allocation matrix A ij ∈ {0, 1} is introduced, the column vector of the matrix represents the UAV, and the row vector represents the sub-region (task region). By using common optimization tools such as simple greedy iteration or 0-1 linear programming, the sum i ∑ j A ij Δ ij is minimized as the target, and the allocation matrix A ij is solved, and the algorithm recalculates ψ j as soon as the power or task progress is updated, ensuring that high-load areas are preferentially allocated to high-capability UAVs, and low-power UAVs fly to light-load areas in time.

[0073] In addition, the dynamic adjustment of the task region allocation satisfies the following constraint conditions:

[0074]

[0075] wherein, m, n respectively represent the number of UAVs and the number of hull task regions, L(I j ) represents the load of the UAV I j , and L(R i ) represents the task load of the task region R i . The formula shows that the total load of all task regions must be equal to the total load of all UAVs.

[0076] Specifically, the region allocation can be divided into the following steps:

[0077] S21, region division: input the corrosion prediction matrix C i,j and the three-dimensional grid G. Use the DARP algorithm to divide n task regions R1, R2, …, R n ;

[0078] S22, calculate the task load function:

[0079] L(R i ) = aA(R i ) + yP(R i )

[0080] S23, calculate the capability vector and the comprehensive capability coefficient: design the flight capability vector of the jth UAV:

[0081] C j = (E j , V j , M j )

[0082] where E j is the percentage of remaining power, v j is the flight speed of the UAV, and M j is the maximum payload of the UAV. After normalizing the capability vector, the comprehensive capability coefficient is obtained:

[0083] ψ j = a1E j + a2V j + a3M j

[0084] where,

[0085] S24, solve the allocation matrix and dynamically reallocate: construct the capability-load matching degree function according to the comprehensive capability coefficient and the load function:

[0086] Δ ij = |ψ j - L(R i )|

[0087] By using common optimization tools such as simple greedy iteration or 0-1 linear programming to minimize i ∑ j A ij Δ ij is the target, solve the allocation matrix A ij , the algorithm recalculates ψ j when receiving power or task progress updates, ensuring that high-load areas are preferentially allocated to high-capability UAVs, and low-power UAVs fly to low-load areas in time.

[0088] The DARP algorithm has the ability to dynamically adjust. When real-time detection data feedback is received, the algorithm can reallocate task areas to ensure that UAVs preferentially cover new corrosion hotspot areas.

[0089] S3, path planning and optimization: based on the assigned task area, the improved A* algorithm is adopted, combined with the priority of the task area to plan the path of the UAV, and the path is smoothed and optimized by the bidirectional Floyd algorithm.

[0090] After completing the task area division, the next step is to plan the optimal path for each UAV, ensuring that the UAV cluster can complete the corrosion area detection with high efficiency. In this embodiment, the improved A* algorithm is used to optimize the path. The A* algorithm mainly evaluates each search point in the target space and selects the optimal prediction point, then uses heuristic search to find the next optimal prediction point until the target position is reached, as shown in the process of Figure 3 The standard cost function of the A* algorithm is defined as:

[0091] f(n) = g(n) + h(n)

[0092] Where f(n) is the comprehensive cost distance value, g(n) is the actual shortest cost distance from the starting node to the target node, h(n) is the heuristic function, representing the cost distance from the current node to the target node.

[0093] However, in the actual ship corrosion detection task, the A* algorithm cannot consider the priority of the ship corrosion area, nor can it dynamically adjust the path planning. Therefore, this embodiment uses an improved A* algorithm, which introduces the priority weight of the corrosion area based on the standard cost function formula, so that the areas with more serious corrosion of the ship can be detected preferentially in the actual task. Specifically, the cost function of the improved A* algorithm is:

[0094]

[0095] Where f(n) is the comprehensive cost distance value, g(n) is the actual shortest cost distance from the starting node to the target node, h(n) is the heuristic function, representing the cost distance from the current node to the target node, l is the distance from the current position to the target position, L is the distance from the starting point to the target point, λ is the coefficient controlling the influence of the task area priority on path planning, and W(n) is the priority weight determined based on the task area priority.

[0096] In the improved cost function, the traditional actual cost g(n) is retained. However, since h(n) uses the Euclidean distance for value, the actual distance from the current node to the target node is always not greater than the value of h(n). The calculation formula of the Euclidean distance is as follows:

[0097]

[0098] Where (x i ,y i ) is the starting point, (xg , y g ) is a target point, (x k , y k ) is a point in n-dimensional space.

[0099] When the current node is far away from the target node, the actual value will be much larger than the estimated value of h(n), resulting in redundant nodes during the search, so the weight of the estimated value h(n) should be increased to improve efficiency; when the current node is close to the target point, the estimated value gradually approaches the actual value. In order to avoid the estimated value being too large, resulting in the failure of searching the optimal path, the weight of the estimated value needs to be reduced, therefore, the embodiment introduces a weight adjustment term of l / L to h(n). In addition, the application also adds a task area priority weight W(n) and a weighting factor λ. Specifically, W(n) is a weight reflecting the severity of the area corrosion, and the weight of the severely corroded area will be higher to ensure that the unmanned aerial vehicle can preferentially select these areas for detection. In addition, λ is a coefficient used to adjust the influence of corrosion priority on path planning. When λ is large, the algorithm will rely more on the priority of the corrosion area, so as to pay more attention to the high corrosion area; when λ is small, the path planning relies more on the heuristic estimation in the traditional A* algorithm.

[0100] On this basis, the application proposes to use a bidirectional Floyd algorithm to smooth and optimize the path planned by the unmanned aerial vehicle, further reducing the inflection points on the path of the unmanned aerial vehicle, so as to improve the detection efficiency of the unmanned aerial vehicle. In the traditional one-way Floyd algorithm, as the space increases, the path generated by the Floyd algorithm will produce a suboptimal solution, and multiple turns and poor smoothness will still occur in the local area. Therefore, the application further increases the reverse Floyd optimization process from the target point to the starting point in the planning, and finally realizes bidirectional optimization. Specifically, the path optimization formula of the bidirectional Floyd algorithm is:

[0101] D[i][j] = min(D[i][j], D[i][k] + D[k][j])

[0102] Where D[i][j] is the shortest path cost from node i to node j, and k is an intermediate node.

[0103] By continuously updating the matrix D, the double Floyd algorithm can smooth the turns in the path, which can make the unmanned aerial vehicle fly as straight as possible during flight, avoid unnecessary flight curvature, and optimize flight efficiency.

[0104] Specifically, the specific implementation steps of the bidirectional Floyd algorithm are as follows:

[0105] First, a weighted graph G = (V, E) is constructed using discrete grid points and edges, with a given set of starting points S and ending points T. Then, during the forward Floyd process, a full-source shortest path algorithm is run once on the original graph G to obtain the distance matrix D. + With the precursor matrix P + Record the shortest cost from any node to all other nodes. Then, for the reverse Floyd graph, invert all edges to obtain the reverse graph G. R In the reverse graph g R Execute Floyd again to get D. - and P - Record the shortest cost from any node to the destination. Next, perform path concatenation. For each group (s, t) ∈ S × T, scan all nodes k ∈ V and select the transit node with the minimum cost. Use P + Backtracking s→k * Then use P - Backtracking k * →t (reverse direction) and splice to form the complete path. Finally, the trajectory is optimized and updated, collinear or small-angle inflection points are deleted, and arcs / B-splines are inserted if necessary to make the curvature of the turning point meet the flight constraints.

[0106] In actual inspection missions, the corrosion areas on the hull surface often change dynamically over time. Therefore, it is necessary to dynamically adjust the priority of the mission area and modify the flight path in a timely manner based on real-time feedback data. However, mission priority is not only based on the severity of corrosion, but also on flight costs. Therefore, the mission area priority is expressed as follows:

[0107] P i =αS corr (R i )+βC path (R i )

[0108] Among them, P i For task region R i priority, S corr (R i ) is region R i The severity of corrosion is expressed as corrosion area or corrosion density, C path (R i ) represents the path cost of the UAV from its current location to the mission area, expressed as flight distance or flight time, with α and β as adjustment coefficients. By adjusting coefficients α and β, the priority of corroded areas and the impact of path cost can be flexibly controlled, ensuring that severely corroded areas are covered first, while avoiding unnecessary flight redundancy.

[0109] S4, dynamic priority adjustment and path modification: in the process of hull corrosion detection, the hull corrosion information fed back by the unmanned aerial vehicle is updated in real time, the priority of the task area is dynamically adjusted, and the path of the unmanned aerial vehicle is modified.

[0110] With the dynamic change of the corrosion area, the unmanned aerial vehicle must modify the flight path according to the new priority, and ensure that the area with high priority can be detected in time. The path modification formula is as follows:

[0111] C new [i][j] = min(C old [i][j], C adjusted [i][j])

[0112] Wherein, C old [i][j] is the original path cost; C new [i][j] is the modified path cost; C adjusted [i][j] is the path cost determined according to the new task area priority. The mechanism of dynamic priority adjustment and path modification greatly improves the flexibility and efficiency of hull corrosion detection in practical application. At the same time, the redundant flight of the unmanned aerial vehicle cluster is avoided, and the flight time and energy loss of the unmanned aerial vehicle cluster are reduced.

[0113] Figure 4 The path optimization and adjustment process of the application is shown, which includes the following steps:

[0114] Initialize the map and parameters;

[0115] Select the calculation node;

[0116] Calculate the task area priority weight (i.e. corrosion area weight);

[0117] Calculate the node cost;

[0118] Path smoothing;

[0119] Update the node;

[0120] According to the obtained new corrosion information, update the corrosion area, if the corrosion area is updated, real-time dynamic feedback and update the task area priority weight, and then carry out the next round of path adjustment; otherwise

[0121] Determine whether the target node is reached, if yes, output the optimal path, end the path planning, otherwise, return to select the next calculation node.

[0122] S5, multi-UAV coordination and conflict management: in the process of UAV cluster performing tasks, real-time detection of UAV path conflict and flight time conflict, if the task area of two UAVs overlap, which will lead to efficiency decline and waste of resources. Therefore, the flight trajectory of UAV cluster needs to be monitored, when the conflict is detected, the path is adjusted to ensure that the path does not cross and minimize flight time and energy consumption. Specifically, in this embodiment, time window and minimum distance are used for judgment. After the path planning of UAV is completed, the path will be discretized into continuous space-time points. In the process of path execution, the flight trajectories of all UAVs are compared periodically, if the spatial distance of any two UAVs is less than the preset safety threshold d min in the same time slice, it is considered that there is a flight conflict. When the conflict is detected, according to the priority of the task area corresponding to the path, the path corresponding to the high priority task is preferentially reserved. And for the low priority UAV path, time offset processing is performed, that is, the path is shifted backward by a time window as a whole, so as to avoid the conflict area at the same time. In addition, if there is still conflict after time adjustment, the local path points are offset, the flight path is fine-tuned in the safe direction within the range of the original task area, so as to pull apart the minimum distance.

[0123] The above describes the preferred embodiments of the application in detail. It should be understood that those skilled in the art can make many modifications and changes without creative labor according to the concept of the application. Therefore, any technical solution obtained by logical analysis, reasoning or limited experiment on the basis of the prior art according to the concept of the application shall be within the protection scope determined by the claims.

Claims

1. A dynamic path planning method for a UAV cluster for hull corrosion detection, characterized in that, The method comprises the following steps: Region division initialization: scanning the hull surface with high-precision sensors to generate a three-dimensional model, and dividing the corrosion region based on a corrosion prediction model; Region allocation: based on the divided corrosion region, dynamically adjusting the task region allocation through the DARP algorithm, and allocating a task region to each UAV, and calculating the task region priority; Path planning and optimization: based on the allocated task region, using an improved A* algorithm to plan the UAV path in combination with the task region priority, and smoothing and optimizing the path through the bidirectional Floyd algorithm; Dynamic priority adjustment and path modification: in the process of hull corrosion detection, real-time update of the hull corrosion information fed back by the UAV, dynamic adjustment of the task region priority, and modification of the UAV path; The dynamic adjustment of the task region allocation through the DARP algorithm specifically comprises: Partitioning by DARP algorithm n Task Area R 1, R 2, ..., R n and assigned to m A drone is deployed, and a task load function is determined based on the corrosion area and density of the region. The drone is then assigned based on this task load function, which is expressed as follows: wherein denotes the task area of the task load, is the area of the task area , is the corrosion distribution density of the task area , and are the respective weight factors.

2. The dynamic path planning method for the UAV cluster for hull corrosion detection according to claim 1, characterized in that, The division of the corrosion region based on the corrosion prediction model specifically comprises: preliminary analysis of the corrosion region through the corrosion prediction model to generate a corrosion prediction matrix: wherein, represents the corrosion prediction value of different regions of the ship body.

3. The dynamic path planning method for the UAV cluster for hull corrosion detection according to claim 1, characterized in that, The dynamic adjustment of the task region allocation further comprises: dynamically adjusting the allocation of the task region according to the flight capability of the UAV to ensure that the task load of each UAV is appropriate, and the flight capability of the UAV is determined by the following method: Definition 1 The flight capability vector of the drone is: wherein, is a percentage of the remaining power of the drone, is a flight speed of the drone, is a maximum payload of the drone; The flight ability vector is normalized to obtain a comprehensive ability coefficient to represent the flight ability: wherein is a weight, .

4. The dynamic path planning method for the UAV cluster for hull corrosion detection according to claim 1, characterized in that, The dynamic adjustment of the task region allocation satisfies the following constraint conditions: wherein, m , n represent the number of UAVs and the number of hull mission areas, respectively, denotes the payload of a UAV , denotes the mission payload of a mission area .

5. The dynamic path planning method for the UAV cluster for hull corrosion detection according to claim 1, wherein, The cost function of the improved A* algorithm is: wherein, is a comprehensive cost distance value, is an actual shortest cost distance from the start node to the target node, is a heuristic function, representing the cost distance from the current node to the target node, l is the distance from the current position to the target position, L is the distance from the start point to the target point, is a coefficient for controlling the influence of the task area priority on path planning, is a priority weight determined based on the task area priority.

6. The dynamic path planning method for the UAV cluster for hull corrosion detection according to claim 1, characterized in that, The path optimization formula of the bidirectional Floyd algorithm is: wherein, D [ i ][ j ] is the shortest path cost from node i to node j , k is an intermediate node; by continually updating the shortest path cost matrix D to smooth out turns in the path.

7. The dynamic path planning method for the UAV cluster for hull corrosion detection according to claim 1, wherein, The task region priority is represented as: wherein, is the priority of the task area , is the corrosion severity of the area , is the path cost of the UAV from the current location to the task area, expressed in terms of flight distance or flight time, and is the adjustment coefficient.

8. The dynamic path planning method for the UAV cluster for hull corrosion detection according to claim 1, characterized in that, In the dynamic priority adjustment and path modification, the path modification formula is as follows: wherein, is the original path cost; is the modified path cost; is the path cost determined according to the new task area priority.

9. The dynamic path planning method for the UAV cluster for hull corrosion detection according to claim 1, wherein, The method further comprises: Multi-UAV cooperation and conflict management: in the process of the UAV cluster performing the task, real-time detection of UAV path conflicts and flight time conflicts, adjustment of the path when a conflict is detected to ensure that the path does not intersect and minimizes the flight time and energy consumption.

Images