Distributed multi-AUV underwater area detection method based on anti-cluster game

Abstract

The present application belongs to the field of underwater terrain detection technology, and provides a distributed multi-AUV underwater area detection method based on anti-clustering game, and performs the following steps for any one of the multiple AUVs: underwater terrain detection based on the current position of the AUV and the effective coverage distance range, and self-updates the information map it carries; mutual update of the information map carried by the AUV and other AUVs in its neighborhood; execution of the best dynamic response algorithm based on local game to determine the position of the next target point of the AUV and other AUVs in its neighborhood; determination of the target velocity vector of the AUV at the next moment based on the anti-clustering strategy; update the current moment and return to execute the first step until the detection of the underwater area is completed or the end moment is reached. The method provided by the present application can effectively improve the coverage efficiency of multiple AUVs in distributed exploration of underwater areas.

Description

Technical Field

[0001] The present application belongs to the technical field of underwater terrain detection, and specifically provides a distributed multi-AUV underwater area detection method based on anti-cluster game. Background Art

[0002] With the continuous advancement of marine science and technology, autonomous underwater vehicles (AUVs) have been widely used in many fields such as marine resource exploration, marine environment monitoring, and seabed topography mapping. Compared with manned underwater vehicles, AUVs have higher work efficiency, better maneuverability, and longer endurance, and can perform various detection, measurement, and sampling tasks in complex underwater environments.

[0003] In actual AUV applications, multiple AUVs are often required to work together to cover a larger area of ​​sea. How to reasonably plan and control the motion trajectories of multiple AUVs to achieve efficient coverage of the target sea area is an important research topic. Traditional methods are mainly based on centralized control, where the ground station uniformly plans and dispatches all AUVs. This method requires AUVs to maintain stable communication with the ground station. Limited by the low bandwidth and high latency characteristics of the underwater acoustic channel, it is difficult to achieve large-scale, multi-node real-time coverage control. In view of the limitations of centralized control, distributed multi-AUV coverage control has received widespread attention in recent years. Under the distributed control framework, each AUV independently plans its motion path based on the local information it obtains, realizes regional division and conflict avoidance through information interaction with neighboring AUVs, and finally completes autonomous coverage of the target sea area. Compared with centralized control, distributed control overcomes the dependence on ground stations and demonstrates better robustness, scalability and fault tolerance.

[0004] However, there are the following challenges in the distributed detection of underwater unknown areas by multiple AUVs: multiple AUVs with limited communication radius are deployed in an unknown underwater environment. At the beginning of the mission, the AUVs know nothing about all the information in the area, including the location and distribution of obstacles. Each AUV must gradually obtain environmental information during the coverage process and adjust its path planning in real time to ensure efficient completion of the full coverage mission. Since each AUV cannot obtain the location of other AUVs or details of the surrounding environment at the beginning, they need to cope with the challenges brought by dynamic path planning and unknown environments through autonomous detection, coordinated cooperation, and rational use of resources. During the detection process, they need to enter the communication radius with each other to ensure that different AUVs can exchange and synchronize information to achieve distributed collaborative detection, and they must prevent multiple AUVs from being too dense and colliding. In addition, excessive concentration of AUVs in certain areas may lead to redundant operations, thereby reducing resource utilization efficiency and delaying the coverage progress of the entire area. Summary of the invention

[0005] The present application provides a distributed multi-AUV underwater area detection method based on anti-cluster game, which is used to control several AUVs to detect underwater areas. To conduct detection, for any AUV, perform the following steps:

[0006] S1, performing underwater terrain detection based on the current position of the AUV and the effective coverage distance range, and self-updating the information map carried by the AUV, wherein the information map is used to represent the situation of the underwater area being detected;

[0007] S2, mutually updating the information graphs carried by the AUV and other AUVs in its neighborhood;

[0008] S3, executing the optimal dynamic response algorithm based on local game for the AUV and other AUVs in its neighborhood to determine the position of the next target point of the AUV and other AUVs in its neighborhood;

[0009] S4, determining the target velocity vector of the AUV at the next moment based on the anti-clustering strategy;

[0010] S5, update the current time and return to execute step S1 until the detection of the underwater area is completed or the end time is reached.

[0011] Furthermore, for any AUVs self-update the information graph they carry based on the following formula:

[0012] ,

[0013] in, , are the AUV at the current moment. and the previous moment Carrying infographics, For underwater areas Each position in is the position of the AUV at the current moment, is the effective coverage distance range of AUV.

[0014] Furthermore, for any For each AUV, the information graph carried by the AUV and other AUVs in its neighborhood is updated based on the following formula:

[0015] ,

[0016] in, is the set of other AUVs in the neighborhood of the AUV, for The Information chart carried by AUV, To find the maximum value function.

[0017] Preferably, step S3 further comprises the following steps:

[0018] S31, construct the AUV and all AUVs in its neighborhood as a local set ,in, For local collection The AUVs, For local collection The total number of AUVs in;

[0019] S32, initialization of iteration times and local collections The detection strategies of each AUV in the

[0020] S33, select local set Any one of ;

[0021] S34 is determined based on the following formula The next destination :

[0022] ,

[0023] in, for The chosen detection strategy, for The detection strategy set, For local collection Except Other AUV joint detection strategies, for The utility function of To find the maximum index function;

[0024] S35, replace with a new one Repeat step S34 until the local set is traversed. All AUVs in;

[0025] S36, determine the local set Whether all AUVs in have reached Nash equilibrium, if yes, then end step S3, if not, then return to execute step S33.

[0026] Preferably, for a local set Any one of , whose utility function Determined based on the following formula:

[0027] ,

[0028] in, is the information graph evaluation function, is the motion consumption-position conflict evaluation function, for The infographic carried in the current moment, For location The time interval from the last detection time to the current time, for The current position To location The distance decay function is is the target distance perception threshold, is the spatial attenuation factor, For As the center of the circle, is the proportion of locations in the circular area with a radius that have not been detected by any AUV, is the weight coefficient of exercise consumption, is the weight coefficient of target position conflict, For local collections except Outside AUVs, for The position of the next target point determined.

[0029] Preferably, for any AUV, when any of the following conditions is met, execute step S3:

[0030] The AUV reaches its target point; the change between the number of AUVs existing in the neighborhood of the AUV at the current moment and the number of AUVs existing at the previous target point update time exceeds a preset change threshold; the ratio of the area where the information graph carried by the AUV changes before and after the mutual update to the total area exceeds a preset ratio threshold.

[0031] Preferably, any The AUV determines its target velocity vector at the next moment based on the following formula :

[0032] ,

[0033] in, is the resultant force acting on the AUV, for The proportionality coefficient, , , are the anti-cluster repulsion, obstacle repulsion and target gravity of the AUV, is the repulsive potential function, is the spatial gradient operator, is the set of other AUVs in the neighborhood of the AUV, For the AUV and The The distance between the AUVs, is the collision range of the AUV, is the set of obstacles in the neighborhood of the AUV, for The The location of the obstacle, is the obstacle collision range, , are the position of the next target point of the AUV and its position at the current moment, respectively. is the gravitational coefficient of the target point, is the velocity vector of the AUV at the current moment, for The inertia coefficient of .

[0034] The present application provides a distributed multi-AUV underwater area detection method based on anti-clustering game, which comprehensively considers the impact of underwater communication distance limitation on distributed AUV detection, and realizes the sharing of local exploration history through self-update and mutual update of information graph; within the communication range, each AUV only exchanges information with AUVs in its neighborhood, and iteratively adjusts the target position through the best response method. This local game model conforms to the characteristics of underwater communication, and can significantly reduce the computational complexity and accelerate the convergence speed. At the same time, the utility function of the local dynamic game is constructed on the basis of comprehensively considering the detection history, distance attenuation, detection degree, motion consumption and target point overlap conflict of each AUV itself and other AUVs at each position, so that the target point selection is more reasonable, and the global efficiency of distributed detection is effectively improved; in addition, the movement of each AUV is driven by the anti-clustering strategy, so as to avoid the collision risk caused by the excessive concentration of AUV distribution. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 A top view schematic diagram of distributed detection of underwater areas by multiple AUVs;

[0036] Figure 2 A schematic side view of distributed detection of an underwater area by multiple AUVs;

[0037] Figure 3 A schematic flow chart of a distributed multi-AUV underwater area detection method based on anti-cluster game provided according to an embodiment of the present application;

[0038] Figure 4 In a specific embodiment, a schematic diagram of the positions of AUV1 and AUV2 when detecting the underwater area to 0.1 seconds;

[0039] Figure 5 is a schematic diagram of an information graph carried by the AUV1 when detecting an underwater area to 0.1 seconds in a specific embodiment;

[0040] Figure 6 is a schematic diagram of an information graph carried by the AUV2 when detecting an underwater area to 0.1 seconds in a specific embodiment;

[0041] Figure 7 In a specific embodiment, a schematic diagram of the positions of AUV1 and AUV2 when detecting the underwater area for 2.5 seconds;

[0042] Figure 8 is a schematic diagram of an information graph carried by AUV1 when detecting an underwater area for 2.5 seconds in a specific embodiment;

[0043] Fig. 9is a schematic diagram of an information graph carried by AUV2 when detecting an underwater area for 2.5 seconds in a specific embodiment;

[0044] Fig.10 In a specific embodiment, a schematic diagram of the positions of AUV1 and AUV2 when detecting the underwater area for 7.7 seconds;

[0045] Fig.11 is a schematic diagram of an information graph carried by AUV1 when detecting an underwater area for 7.7 seconds in a specific embodiment;

[0046] Fig.12 is a schematic diagram of an information graph carried by AUV2 when detecting an underwater area for 7.7 seconds in a specific embodiment;

[0047] Fig.13 The following is a schematic diagram of the coverage of an underwater area by executing the method provided in the embodiment of the present application in a specific embodiment. DETAILED DESCRIPTION

[0048] Hereinafter, the present application will be further described based on preferred embodiments with reference to the accompanying drawings.

[0049] <1. Application scenarios and nouns and terminology>:

[0050] In order to facilitate a clear understanding of the technical solutions of the embodiments of the present application, the application scenarios of the methods provided in the embodiments of the present application and related nouns and terms are first explained.

[0051] 1.1) Distributed AUV underwater terrain detection:

[0052] The method provided in this application is used to control multiple AUVs (Autonomous Underwater Vehicles) to navigate underwater areas. For more information, please refer to Figure 1 and Figure 2 Under the control of their respective operating systems, AUVs navigate underwater through power systems (generally composed of motors and propellers), steering systems (such as steering rudders), etc., and use underwater camera modules or multi-beam sonar and other detection devices to monitor the underwater area during navigation. Conduct optical and acoustic detection, and finally complete the underwater area After detection, its two-dimensional or three-dimensional terrain data is obtained.

[0053] 1.2) Target point and target velocity vector:

[0054] Each AUV is in the underwater area In the process of detection, it is generally set by setting the target point (such as Figure 1 The AUV control unit determines its target velocity vector (e.g., Figure 1 The power system and steering system are then controlled to navigate at the set target speed vector.

[0055] 1.3) Detected range and information diagram:

[0056] Obviously, at any moment, each AUV will take its location as the center of the circle and the effective coverage distance range as the radius, forming a circle like Figure 1 The effective coverage area shown by the red dotted circle in the middle, as the AUV continues to navigate, the envelope formed by the red dotted circles corresponding to each point in its navigation trajectory constitutes the detection range of each AUV.

[0057] In order to avoid repeated detection of the detected range as much as possible, it is necessary to clearly mark the detected range of the AUV and the area that has not been detected, so as to better plan the target point of the AUV. To this end, in an embodiment of the present application, each AUV records the range it has detected and the time when each position in the range was last detected through the information map it carries.

[0058] In some specific embodiments, the information map can be constructed and represent the underwater area in the following manner: Detection situation: underwater area Rasterize it so that Various locations within Mapped to each grid point in the information graph, when the value of a grid point is 0, it represents the position corresponding to the grid point has not been detected by AUV; accordingly, when a position When an AUV detects a location at a certain time after the start of detection, the AUV sets the value of the corresponding grid point on the information map it carries to the value of that time. Obviously, since the same location may be detected repeatedly, a grid point on the information map may be updated multiple times, and each time it is updated, the time of its most recent detection will be stored. In this way, each grid point in the information map carried by an AUV records whether the AUV has been detected and the specific time of its most recent detection.

[0059] In some embodiments, the information map can be constructed and stored in the form of a two-dimensional table. When the underwater area being explored is large or the grid density is high, the information map can be displayed in the form of a heat map or cloud map, using the depth of color to indicate the latest time a location was detected.

[0060] 1.4) Neighborhood:

[0061] In order to efficiently detect underwater areas, multiple AUVs need to work together. Underwater communication generally uses hydroacoustic signals. The signal strength decays rapidly with the increase of propagation distance. Its communication range (such as Figure 1 The blue dotted circle in the middle is much smaller than the wireless communication method on the ground or in the air. When the detection area is large, it is impossible to achieve centralized control of all AUVs through a main control unit. Each AUV can only communicate with other AUVs within its communication range to perform distributed collaborative control. In the embodiment of the present application, for this distributed collaborative control method, the communication range of any AUV is regarded as its neighborhood. Obviously, the AUV can only communicate and exchange data with other AUVs in its neighborhood.

[0062] 1.5) Optimal dynamic response algorithm:

[0063] The best dynamic response algorithm is a strategy optimization algorithm in game theory. It optimizes the global strategy by iteratively analyzing and updating the strategies of all participants in the game. The algorithm generally includes the following steps (assuming that there are Participants):

[0064] First, the strategy set of each participant is determined.

[0065] Specifically, for the ( ) participants, whose strategy set All the available strategies of the participant Composition, that is , accordingly, the joint strategy of all participants can be expressed as a vector , which represents the vector of strategies chosen by all participants. Obviously, this vector The Cartesian product of the strategy sets belonging to all participants satisfies the following equation: .

[0066] Then, all participants select an initial strategy from their strategy set.

[0067] After completing the above initialization operations, in each round, select one participant in turn , according to which the participant can currently The strategies selected in and the joint strategy consisting of the current strategies of all other participants Calculating the utility function , where the utility function represents the joint strategy consisting of the strategies chosen by all other participants In the known case, Participants choose strategies The benefits that can be obtained, then choose your best response strategy (that is, the strategy corresponding to the maximum value of the utility function) ), then replace the next participant and repeat the selection of the best response strategy above until all participants are traversed;

[0068] After the traversal is completed, it is determined whether the strategies selected by all participants have reached the end condition. If so, the algorithm is terminated. If not, a new round of optimal dynamic response iteration is re-executed.

[0069] The end condition can be a preset upper limit of the number of iterations, or the Nash equilibrium condition in game theory can be used. In game theory, Nash equilibrium means that each participant has no motivation to unilaterally change his or her strategy given the strategies of other participants, that is, when the system reaches Nash equilibrium, no participant can further improve the value of his or her own utility function by changing his or her own strategy.

[0070] <2. Challenges faced by multi-AUV collaborative exploration>:

[0071] As mentioned above, when an underwater area is explored collaboratively by multiple AUVs, due to the limitation of underwater communication distance, it is impossible to be uniformly controlled by a main control system. Therefore, how to coordinate multiple AUVs to fully explore the entire underwater area in the shortest possible time faces many challenges:

[0072] 2.1) Repeated detection problem:

[0073] For the same location, it may be detected multiple times by different AUVs, and it may also be repeatedly detected by the same AUV. Obviously, the more times it is repeatedly detected, the lower the efficiency of the detection algorithm. For this reason, it is necessary to realize the sharing of detected and undetected ranges among various AUVs when the underwater communication range is limited and there is no unified control mechanism. At this time, each AUV can fully consider its own and other AUVs' detection history information when determining the motion strategy, and avoid repeated detection of the detected range as much as possible.

[0074] 2.2) Aggregation Problem:

[0075] In addition to the problem of repeated exploration, the uneven distribution of multiple AUVs in different areas will also cause a decrease in global exploration efficiency. In addition, when there are too many AUVs in a certain area, or the distance between different AUVs is too small, there may be a risk of collision.

[0076] 2.3) Target point update strategy and timing:

[0077] Compared with the detection strategy of a single AUV, when multiple AUVs perform collaborative detection, the target point selection strategy needs to comprehensively consider its own detection situation and the detection situation of other AUVs. As mentioned above, due to the limitation of communication range, an AUV can only communicate and exchange information with the AUVs in its neighborhood. As each AUV moves, the number and number of AUVs in its neighborhood are constantly changing. If the target point is updated too slowly, the AUV may not be able to adjust its own detection strategy in time according to the detection strategy of its neighboring AUVs, resulting in a decrease in global detection efficiency. If the target point is updated too quickly, it will not only increase the consumption of computing resources, but also cause additional maneuvers due to frequent changes in the target point, greatly increasing the power loss of the AUV.

[0078] <3. Implementation Methods of the Application>:

[0079] This application provides a distributed multi-AUV underwater area detection method based on anti-cluster game through an embodiment, which controls several AUVs to detect underwater areas by optimizing the distributed coordination mechanism. Comprehensive and efficient detection is carried out, wherein the number of AUVs can be determined by comprehensively considering factors such as navigation speed, endurance time, and the area of ​​underwater area to be detected. When the number of AUVs is two or more, the method provided in this application can be executed on any one AUV.

[0080] Specifically, for any AUV, it executes Figure 3 Steps shown:

[0081] S1, performing underwater terrain detection based on the current position of the AUV and the effective coverage distance range, and self-updating the information map carried by the AUV, wherein the information map is used to represent the situation of the underwater area being detected;

[0082] S2, mutually updating the information graphs carried by the AUV and other AUVs in its neighborhood;

[0083] S3, executing the optimal dynamic response algorithm based on local game for the AUV and other AUVs in its neighborhood to determine the position of the next target point of the AUV and other AUVs in its neighborhood;

[0084] S4, determining the target velocity vector of the AUV at the next moment based on the anti-clustering strategy;

[0085] S5, update the current time and return to execute step S1 until the detection of the underwater area is completed or the end time is reached.

[0086] Obviously, when the effective coverage area of ​​each AUV does not overlap with the underwater area When the intersection occurs, or before the detection of the underwater area begins, the underwater area carried by each AUV In the information diagram, the value of each grid point is 0; any AUV starts to enter the underwater area in the effective coverage area After the detection time starts, the above steps S1 to S5 are cyclically executed until the underwater area is completed. The detection is completed or a preset end time is reached (the end time can be predetermined according to factors such as the endurance of each AUV or the detection plan).

[0087] The specific implementation process of the above steps S1 to S5 is described in detail below with reference to the accompanying drawings.

[0088] 3.1) Underwater area detection and self-updating of information maps:

[0089] For any AUV (without loss of generality, represents the number of the AUV among all AUVs participating in the detection), and executes step S1 to The optical and acoustic detection methods used to detect the location and underwater terrain within the effective coverage area have been introduced in detail in the previous article and will not be repeated here.

[0090] In this When an AUV detects an underwater area, the information map it carries will be updated in real time. The self-update of is performed by formula (1):

[0091] (1)

[0092] in, For the current moment, For the previous moment, , are the AUV at the current moment. and the previous moment Carrying infographics, For underwater areas Each position in (for the gridded underwater area , as mentioned above, That is, each grid point), is the position of the AUV at the current moment, is the effective coverage distance range of AUV, and (1) represents the current time , within the effective coverage area of ​​the AUV , the value of its information graph is updated to , which is not within its effective coverage area , the value of its information graph remains the same as the previous moment (i.e. ) when .

[0093] Figure 4 It shows the positions of the two AUVs (AUV1 and AUV2) in a specific embodiment when performing collaborative detection of an underwater area. The green circle and the blue circle respectively represent the positions of AUV1 and AUV2 in the underwater area, and the green star and the blue star respectively represent the next target point positions of AUV1 and AUV2. Figure 5 and Figure 6 The information graphs carried by AUV1 and AUV2 at 0.1 seconds are shown respectively.

[0094] like Figures 4 to 6 As shown in Figure 1, at 0.1 seconds, the information graphs carried by the two AUVs only have the positions of the two AUVs as the centers of the circles. The value of the information graph is not zero within a circle with a radius of . As the AUV continues to sail, its effective coverage area is constantly sweeping the underwater area as its position changes. For any position, its value in the information graph carried by an AUV represents the last time it was in the effective coverage area of ​​the AUV. When it leaves the effective coverage area of ​​the AUV, its value in the information graph remains unchanged until it is swept again by the effective coverage area of ​​the AUV or updated by the mutual update operation described below.

[0095] 3.2) Sharing of detection information between multiple AUVs:

[0096] In the process of collaborative detection by multiple AUVs, in order to avoid the decrease in global efficiency caused by repeated detection, each AUV not only records its own detected range through an information map, but also needs to share the information map it carries with other AUVs, so that each AUV can determine a reasonable target based on the detected range completed by itself and other AUVs. In the embodiment of the present application, the information sharing of any AUV with other AUVs is carried out by mutual updating of the information map in step S2.

[0097] As mentioned above, due to the underwater communication distance limitation, the mutual update of any AUV can only be carried out within the neighborhood determined by its communication range. For each AUV, the information graph carried by the AUV and other AUVs in its neighborhood can be mutually updated based on formula (2):

[0098] (2)

[0099] in, is the set of other AUVs in the neighborhood of the AUV, for The Information chart carried by AUV, To find the maximum value function.

[0100] Obviously, in the During the mutual update of the information graph by each AUV, the other AUVs in its neighborhood also realize the mutual update of the information graphs they carry. After the mutual update is completed, the information graphs carried by all AUVs in the neighborhood are synchronized, and the values ​​at each position record the time when the position was most recently detected by any AUV in the neighborhood.

[0101] Figure 7 Shows Figure 4 The positions of the two AUVs in the figure at 2.5 seconds of detection. Figure 8 , Fig. 9 The information graphs carried by AUV1 and AUV2 at 2.5 seconds are shown respectively; Fig.10 Shows Figure 4 The positions of the two AUVs in the image at 7.7 seconds. Correspondingly, Fig.11 , Fig.12 The information graphs carried by AUV1 and AUV2 at 7.7 seconds are shown respectively.

[0102] like Figures 7 to 9 As shown in FIG. 2 , at 2.5 seconds, the two AUVs have not yet entered the communication range, so AUV1 and AUV2 still independently update the information map they carry; Figures 10 to 12 As shown in Figure 2, at 7.7 seconds, the two AUVs are close enough to communicate, that is, they enter each other's neighborhood. At this time, the information graphs carried by the two AUVs can be synchronized by mutual updating. Fig.11 and Fig.12 The displayed information graphs become the same, that is, AUV1 and AUV2 can obtain their own and each other's detection history information of the underwater area through the mutual update of the information graphs. Fig.10It can also be seen that at 7.7 seconds, both AUVs re-determined their next target points.

[0103] 3.3) Optimal dynamic response algorithm based on local game:

[0104] Step S3 is used to determine the target point for each AUV to explore the underwater area. When multiple AUVs conduct terrain exploration in a distributed underwater area, the control system of each AUV determines its own target point through a certain detection strategy, and drives its power system and steering system to navigate to the target point. The selection of the target point by each AUV needs to comprehensively consider the detection history and target point conditions of itself and other AUVs to improve the global detection efficiency. At the same time, since the update of the target point cannot be performed on all AUVs at the same time, it can only be performed in the local area where they can communicate with each other. Therefore, the update timing of the target point should also be comprehensively evaluated to avoid the waste of computing power and power resources caused by too short target point update intervals, and the overall detection efficiency reduction caused by too long update time.

[0105] In an embodiment of the present application, each AUV selects a target point in step S3 by executing the best dynamic response algorithm based on local game, wherein the local game means that for any AUV, the determination of its next target point is carried out through a game between the AUV and a local AUV set composed of other AUVs in its neighborhood. In some preferred embodiments, step S3 includes the following steps:

[0106] S31, construct the AUV and all AUVs in its neighborhood as a local set ,in, For local collection The AUVs, is the total number of AUVs in the local ensemble.

[0107] S32, initialization of iteration times and local collections Detection strategies of each AUV in .

[0108] S33, select local set Any one of .

[0109] S34, determined based on the following formula (3) The next destination :

[0110] (3)

[0111] (3) In the formula, for The chosen detection strategy, for The detection strategy set, For local collection Except Other AUV joint detection strategies, for The utility function of To find the maximum index function.

[0112] about , , as well as The meaning of has been introduced in detail in the previous terminology section. Obviously, for underwater detection problems, Each selectable detection strategy This represents the underwater area Correspondingly, the detection strategy set of each AUV is .

[0113] S35, replace with a new one Repeat step S34 until the local set is traversed. All AUVs in the

[0114] S36, determine the local set Whether all AUVs in have reached Nash equilibrium, if yes, then end step S3, if not, then return to execute step S33.

[0115] 3.4) Optimization of utility function:

[0116] In the above-mentioned iterative optimization process of the best dynamic response based on local game, whether the utility function is set reasonably determines whether the selection of the next target point is appropriate to a large extent and has a key impact on the overall detection efficiency. In some preferred embodiments of the present application, for any one of the local sets , whose utility function The following formula (4) is adopted:

[0117] (4)

[0118] (4) It consists of two parts as a whole: is an information graph evaluation function, which is used to evaluate the information value of each grid in the information graph; It is the motion consumption-position conflict evaluation function, which is used to evaluate the conflict between the motion consumption and position of the AUV.

[0119] a. Information graph evaluation function:

[0120] Information graph evaluation function is used to evaluate underwater areas Various locations in The detected value is evaluated. In some preferred embodiments, the information graph evaluation function is determined based on the following formula (5):

[0121] (5)

[0122] (5) In the formula, for The information graph carried at the current moment, accordingly, Represents the location The time interval from the last detection time to the current time (because the information diagram Locations are stored in The last detected time information), Indicates that the AUV (i.e. ) at the current time To location The distance decay function is is the target distance perception threshold, is the spatial attenuation factor, It means As the center of the circle, is the proportion of locations not detected by any AUV in a circular area with a radius of The information carried in the figure is As the center of the circle, The detected area in the circular area with a radius of (i.e., the area or number of grids covered by the non-zero position in the information graph) is counted, and then divided by the total area of ​​the circular area (or the total number of grids in the circular area), so as to obtain .

[0123] From the information graph evaluation function constructed by equation (5), we can see that the evaluation of the information graph includes the time factor, distance attenuation factor and detection situation, among which: the time term The larger the The longer a nearby location has not been detected, the greater the utility of driving the AUV toward it; The smaller, the The closer to the AUV target point, the closer the target point is to the boundary of the AUV target point. , correspondingly, the larger the distance attenuation term is, the greater the utility of the AUV moving to that position is. Distance from current location hour, The value of is the largest; Indicates by location Centered on The proportion of undetected locations in a circular area with a radius of [0,1]. The smaller the value is, the larger the undetected area is, which means the higher the utility of AUV moving towards it.

[0124] It can be seen that in the embodiments of the present application, through the detection time information contained in the information map carried by each AUV, the utility of each position can be evaluated from three aspects: time interval, distance attenuation and detection status, so as to help the AUV determine the position of the next target point more reasonably.

[0125] b. Motion consumption-position conflict evaluation function:

[0126] Motion cost-position conflict evaluation function It is used to evaluate the consumption of the AUV's movement when it moves to a target point and the cost of repeated detection with other AUVs that also move to the same location. In some preferred embodiments, the function is determined by formula (6):

[0127] (6)

[0128] in, is the weight coefficient of exercise consumption, is the weight coefficient of the target position conflict, For local collection Medium Outside AUVs, for The position of the next target point is determined. Obviously, the target point is The farther the current position is from the current position, the higher the motion consumption. The closer the target point of an AUV is set to the target point of another AUV, the greater the cost of the conflict between the two.

[0129] From the above analysis, it can be seen that the utility function constructed in this application can comprehensively consider the detection history, distance attenuation, detection degree, motion consumption, and target point overlap conflicts of each AUV and other AUVs at each location during the collaborative detection of multiple AUVs, thereby making the target point selection more reasonable and effectively improving the global efficiency of distributed detection.

[0130] 3.5) Timing of updating the target point:

[0131] During the distributed detection process, each AUV can only conduct the optimal strategy game optimization within the neighborhood through local games and mutual updates of information graphs. As each AUV continues to move, the number of AUVs in the neighborhood will continue to change. Therefore, in addition to the need to re-determine the next target point after an AUV reaches its set target point, when the AUVs in the neighborhood of an AUV change significantly, the original optimal strategy of the AUV may become invalid. At this time, it is necessary to reselect the next target point based on the increase or decrease of AUVs in its neighborhood. At the same time, it should be noted that too frequent updates to the next target point not only require multiple AUVs to perform iterative game operations at the same time, but also cause the AUV to frequently perform maneuvering operations such as turning, which greatly increases the consumption of AUV calculation and movement.

[0132] To this end, in some preferred embodiments, for any When an AUV reaches the target point, or when the change between the number of AUVs existing in the neighborhood of the AUV at the current moment and the number of AUVs existing at the previous target point update moment exceeds a preset change threshold, step S3 is executed.

[0133] The change threshold can be dynamically adjusted according to the number of AUVs in the neighborhood of the AUV at the last target point update. For example, at a certain moment, there are 10 AUVs in the neighborhood of an AUV. After it and the other 9 AUVs in its neighborhood determine the target point through the best dynamic response algorithm, the AUVs entering and leaving its neighborhood are continuously counted. If the number of AUVs in its neighborhood changes to 15 or 5 at a certain moment thereafter, that is, the increase or decrease is 50% of the previous target point update time, the preset change threshold is reached. At this time, continuing to move to the original target point may no longer be the optimal strategy. It is necessary to re-execute the best dynamic response algorithm for each AUV in its neighborhood at this moment to re-determine the position of its next target point. Obviously, if there are 20 AUVs in the neighborhood of the AUV at the last target point update time, the corresponding change threshold will also be adjusted to plus or minus 10.

[0134] In addition to the above two situations, when any one of them occurs, the target point is triggered to update. There is another situation that will make the original target point no longer suitable. Specifically, for any For an AUV, when the other AUVs entering or leaving its neighborhood are relatively close to its detection trajectory, that is, when the areas detected by these AUVs are relatively concentrated, the change of the information graph caused by each mutual update operation is relatively gentle; however, when a certain AUV The AUV whose trajectory is significantly different from the detected area enters the After the AUVs are in the neighborhood (this may happen, for example, when two AUVs traveling in opposite directions gradually approach each other from a distance far away from each other to a point where they can communicate with each other), the information map it carries may contain a large number of The historical detection information of the area that the AUV has never explored will result in the The values ​​of a large number of grid points that were originally 0 in the information map carried by the AUV have changed. In this case, even if the The number of AUVs in the neighborhood of an AUV has only increased by one, and a large amount of information that has not been detected by itself but has been detected by other AUVs will be "injected" into the information map it carries, which will cause a drastic change in the utility value of the originally planned target point. Therefore, it is necessary to re-determine the next target point.

[0135] To this end, in some preferred embodiments of the present application, for any For an AUV, when the ratio of the area where the value of the information map carried by it changes before and after the mutual update to the whole area exceeds a preset ratio threshold, the next target point is re-determined through step S3.

[0136] For example, in a specific embodiment, the ratio threshold may be set to That is, when the information map carried by an AUV has a change in value of more than 25% of the area in its information map before and after mutual updating, it means that another AUV entering the neighborhood of the AUV has brought a large amount of new detection information and may cause a significant change in the detection value of the target point. At this time, a re-evaluation is needed to select a more appropriate target point.

[0137] It should be noted that, on the one hand, the ratio threshold should not be set too small, so as to avoid triggering the replacement of the target point when only a small part of the information map is updated; on the other hand, the ratio threshold should not be set too large, for example, exceeding , it may be difficult to trigger the target point change. In some preferred embodiments, the above ratio threshold can be ~ Choose between.

[0138] 3.6) Nash equilibrium conditions:

[0139] In some specific embodiments, the Nash equilibrium condition is as follows:

[0140] (7)

[0141] in, for Apart from Other optional target point locations, For local collection Except The vector of the positions of the next target points selected by all other AUVs is formed. Formula (7) indicates that when the Nash equilibrium condition is reached, for any , it is impossible to increase the utility function value by changing the position of its next target point.

[0142] <4. AUV motion control based on anti-clustering strategy>:

[0143] After each AUV determines its target point, it can move toward the target point through the power system and the steering system in step S4. In some specific embodiments, the motion control of the AUV is performed based on the anti-clustering strategy. Specifically, for any For an AUV, its target velocity vector at the next moment can be determined based on the following equation (8): :

[0144] (8)

[0145] in, is the resultant force acting on the AUV, for The proportionality coefficient, , , are the anti-cluster repulsion, obstacle repulsion and target gravity of the AUV, is the repulsive potential function, is the spatial gradient operator, is the set of other AUVs in the neighborhood of the AUV, For the AUV and The The distance between the AUVs, is the collision range of the AUV, is the set of obstacles in the neighborhood of the AUV, for The The location of the obstacle, is the obstacle collision range, , are the position of the next target point of the AUV and its position at the current moment, respectively. is the gravitational coefficient of the target point, is the velocity vector of the AUV at the current moment, for The inertia coefficient of .

[0146] The repulsive potential function may be a potential function known to those skilled in the art. For example, in some specific embodiments, for the distance between The repulsive potential function of two points is shown in formula (9):

[0147] (9)

[0148] in, is the range of repulsive force, is the repulsion coefficient.

[0149] From (8), we can see that for any An AUV will determine its movement direction and speed (i.e., target motion vector) under the combined effect of the gravity generated by the target point (including the inertia factor generated by the current speed), the repulsion of other AUVs in the neighborhood, and the repulsion of obstacles, and use this target motion vector to drive the AUV to move toward the target.

[0150] The above is a detailed description of the implementation process of the distributed multi-AUV underwater area detection method based on anti-cluster game provided by this application. Fig.13 FIG. 4 shows how the cumulative coverage changes over time when detecting an underwater area using the method in a preferred embodiment. For comparison, Fig.13 The results of using only the anti-clustering algorithm without game theory and the results of using the random algorithm are also shown. Fig.13 It can be seen that the use of the method of the present application can effectively improve the global efficiency of underwater area detection, thereby achieving full coverage of the underwater area as quickly as possible.

[0151] The above is a detailed introduction to the specific implementation methods of the present application. For those skilled in the art, several improvements and modifications may be made to the present application without departing from the principles of the present application. These improvements and modifications also fall within the scope of protection of the claims of the present application.

Claims

1. A distributed multi-AUV underwater area detection method based on anti-cluster game, used to control several AUVs to detect underwater area X, characterized in that: For any AUV, perform the following steps: S1, performing underwater terrain detection based on the current position of the AUV and the effective coverage distance range, and self-updating the information map carried by the AUV, wherein the information map is used to represent the situation of the underwater area being detected; S2, mutually updating the information graphs carried by the AUV and other AUVs in its neighborhood; S3, executing the optimal dynamic response algorithm based on local game for the AUV and other AUVs in its neighborhood to determine the position of the next target point of the AUV and other AUVs in its neighborhood; S4, determining the target velocity vector of the AUV at the next moment based on the anti-clustering strategy; S5, updating the current time and returning to step S1 until the detection of the underwater area is completed or the end time is reached; Step S3 further comprises the following steps: S31, construct the AUV and all AUVs in its neighborhood as a local set A = {a i },i∈1...I, where, a i is the i-th AUV in the local set A, I is the total number of AUVs in the local set A; S32, initializing the number of iterations and the detection strategy of each AUV in the local set A; S33, select any a in the local set A i ; S34, determine a based on the following formula i The next destination Among them, s i for a i The selected detection strategy, S set,i for a i The detection strategy set, s -i is the local set A except a i The joint detection strategy of other AUVs other than U i (s i ,s -i ) is a i The utility function of , argmax{} is the maximum value index function; S35, replace with a new i And repeat step S34 until all AUVs in the local set A are traversed; S36, determining whether all AUVs in the local set A have reached Nash equilibrium, if so, ending step S3, if not, returning to step S33; For any a in the local set A i , whose utility function U i (s i ,s -i ) is determined based on the following formula: Among them, ξ i (s i ,s -i ) is the information graph evaluation function, C i (s i ,s -i ) is the motion consumption-position conflict evaluation function, G i (x,t) is a i The information graph carried at the current moment, t-Gi(x,t) is the time interval from the last detection time of position x to the current moment, for a i At the current position x i The distance attenuation function to position x, l is the target distance perception threshold, σ is the spatial attenuation factor, ρ(x,r) is the proportion of positions that have not been detected by any AUV in the circular area with x as the center and r as the radius, α1 is the weight coefficient of motion consumption, α2 is the weight coefficient of target position conflict, and a j is the local set A except a i The jth AUV outside for a j The position of the next target point determined; For any m-th AUV, step S3 is performed when any of the following conditions is met: The AUV reaches its target point; The change between the number of AUVs existing in the neighborhood of the AUV at the current moment and the number of AUVs existing at the previous target point update moment exceeds a preset change threshold; The ratio of the area where the value of the information map carried by the AUV changes before and after the mutual update to the entire area exceeds a preset ratio threshold.

2. The distributed multi-AUV underwater area detection method based on anti-cluster game according to claim 1 is characterized in that: For any m-th AUV, the information graph it carries is self-updated based on the following formula: Among them, G m (x,t),G m (x, t-1) are the information graphs carried by the AUV at the current time t and the previous time t-1, respectively. x is the position in the underwater area X, and x m is the current position of the AUV, r s is the effective coverage distance range of AUV.

3. The distributed multi-AUV underwater area detection method based on anti-cluster game according to claim 2 is characterized in that: For any m-th AUV, the information graphs carried by the AUV and other AUVs in its neighborhood are mutually updated based on the following formula: G m (x,t)=G n (x,t)=max{G m (x,t),G n (x,t)},n∈N m , Among them, N m is the set of other AUVs in the neighborhood of the AUV, G n (x,t) is N m The information diagram carried by the nth AUV in , max{} is the maximum value function.

4. The distributed multi-AUV underwater area detection method based on anti-cluster game according to claim 1 is characterized in that: The specific anti-clustering strategy is that any m-th AUV determines its target velocity vector at the next moment based on the following formula: in, is the resultant force on the AUV, k is The proportionality coefficient, are the anti-cluster repulsion, obstacle repulsion and target gravity of the AUV, is the repulsive potential function, is the spatial gradient operator, N m is the set of other AUVs in the neighborhood of the AUV, z mn For the AUV and N m The distance between the nth AUV in the image, d is the collision range of the AUV, is the set of obstacles in the neighborhood of the AUV, for The position of the pth obstacle in , d β is the obstacle collision range, x m are the position of the next target point of the AUV and its position at the current moment, respectively. s is the gravitational coefficient of the target point, v m is the velocity vector of the AUV at the current moment, k v v m The inertia coefficient of .

Images