A hierarchical target search and capture method for unmanned vehicle swarms

Abstract

The invention relates to a hierarchical target search and capture method for an unmanned vehicle cluster, comprising: dividing an unmanned vehicle cluster into a plurality of small groups, each of which comprises a guide vehicle and two member vehicles; in a to-be-searched area, the guide vehicle of each small group performs a target search based on its own position; when no target is found, the guide vehicles of each small group respectively perform incremental environmental node exploration and continue to perform target search: storing respective current positions in an explored stack and moving to the latest node position in a corresponding to-be-explored stack to perform target search until the search target is found or the nodes in the respective corresponding to-be-explored stacks are empty; wherein the to-be-explored stack is a storage space for a plurality of traversable node positions obtained by the guide vehicle performing radial node exploration at the target search position; the explored stack is used to store the node positions that have been reached by the guide vehicles of each small group; when a target is found, the guide vehicle of the small group performs motion trajectory prediction on the target; and the three vehicles of the current small group perform triangular encirclement and capture on the target.

Description

Technical Field

[0001] The present invention relates to the field of unmanned vehicle cluster technology, and in particular to a hierarchical target search and capture method for unmanned vehicle clusters. Background Art

[0002] Unmanned vehicle swarm control methods usually adopt two modes: centralized or fully distributed. Centralized control relies on a central node to coordinate all unmanned units. Although this method is convenient for unified management and scheduling of global tasks, it has obvious single point failure risks and may cause serious communication delay problems in large-scale clusters. The fully distributed control method improves the robustness and flexibility of the system by giving each unmanned unit the ability to make independent decisions, but it often seems incapable of complex task allocation and resource scheduling, especially in tasks that require high coordination, such as target search and roundup.

[0003] In unknown or dynamic environments, traditional static map building methods usually require prior environmental information, which is not feasible for rapidly deployed unmanned vehicle swarms. On the other hand, although random exploration strategies do not require prior knowledge of the environment, the lack of effective path planning may lead to long exploration times and repeated exploration of the same area, reducing exploration efficiency.

[0004] In the field of target tracking, simple tracking algorithms can often only track the target based on its current position, which ignores the target's future movement trend and easily causes the target to escape. Although some complex prediction models can improve tracking accuracy, they usually consume a lot of computing resources, and the complexity of the model will directly affect the accuracy and real-time performance of the prediction. Summary of the invention

[0005] In view of the above analysis, an embodiment of the present invention aims to provide a hierarchical target search and capture method for unmanned vehicle clusters, so as to solve the problems of poor environmental adaptability, low exploration efficiency and inability to capture targets in real time and dynamically when existing unmanned vehicle clusters search for targets in unknown or dynamic environments.

[0006] The purpose of the present invention is mainly achieved through the following technical solutions:

[0007] The present invention provides a hierarchical target search and capture method for unmanned vehicle clusters, comprising the following steps:

[0008] Divide the unmanned vehicle cluster into several groups, each of which includes a leader vehicle and two member vehicles;

[0009] In the area to be searched, each group of guide vehicles searches for targets based on its own position;

[0010] When the target is not found, each group of guide vehicles respectively conducts incremental environment node exploration based on its own position to continue the target search, including: storing each current position in the explored stack and moving to the latest node position in the corresponding to-be-explored stack to search for the target until the search target is found or the node in the corresponding to-be-explored stack is empty; wherein, the to-be-explored stack is a storage space for the guide vehicle to perform radial node exploration at the target search position to obtain several passable node positions in the to-be-searched area; the explored stack is used to store the node positions that each group of guide vehicles has reached;

[0011] When a target is found, the guide vehicle of the group predicts the motion trajectory of the target; based on the predicted motion trajectory, the three vehicles of the current group surround and capture the target in a triangular manner.

[0012] Furthermore, the incremental environment node exploration stores the initial position of each group of guide vehicles into their respective to-be-explored stacks; during the search process:

[0013] Using the latest node position in the to-be-explored stack as the node position for this search and deleting the node position from the to-be-explored stack;

[0014] When the node position does not exist in the explored stack, moving the guide vehicle to the node position for target search, and storing the node position in the explored stack;

[0015] If the search does not find the target, a 360° radial node exploration is performed with the guide vehicle as the center to obtain several traversable node positions in the area to be searched; and the traversable node positions that do not exist in the explored stack or in the to-be-explored stack are stored in the to-be-explored stack;

[0016] The next search is performed until the search target is found or the to-be-explored stack is empty, and the search is terminated.

[0017] Furthermore, several accessible node locations in the area to be searched are obtained, including:

[0018] With the guide vehicle as the center and the front of the vehicle body as 0°, the environment is continuously scanned 360° counterclockwise at preset angles to obtain the distance between the object and the guide vehicle at each angle;

[0019] Constructing an angle distance array based on the angle and the object distance obtained by scanning the angle;

[0020] Based on the angle distance array, obtaining the position of each object around the guide vehicle;

[0021] Subtract the preset safety distance from the position of each object to obtain the corresponding passable node position.

[0022] Furthermore, when the distance between the object and the guide vehicle at each angle is obtained by continuous scanning, when the distance is greater than a preset maximum distance, the preset maximum distance is used as the distance between the object and the guide vehicle at the current angle.

[0023] Furthermore, obtaining the positions of various objects around the guide vehicle based on the angle distance array includes:

[0024] Traversing the angle distance array and reading the data of the angle distance array;

[0025] Based on the position of the guide vehicle, each angle and distance in the angle distance array, the position of each object around the guide vehicle is obtained; wherein, when the difference between the distances of consecutive angles in the angle distance array is within the distance fluctuation threshold, the objects in this continuous angle range are regarded as continuous objects, the midpoint of this continuous angle range is taken as the angle of the object relative to the guide vehicle, and the shortest distance in the continuous angle range is taken as the distance to the guide vehicle.

[0026] Furthermore, based on the position of the guide vehicle, each angle and distance in the angle distance array, the position of each object around the guide vehicle is obtained using the following formula:

[0027]

[0028]

[0029] Among them, (x i ,y i ) represents the position coordinates of the i-th object around the guide vehicle; (x guide ,y guide ) represents the position coordinates of the guide vehicle; θ i represents the angle of the i-th object around the guide vehicle relative to the guide vehicle; d i Represents the distance between the i-th object around the guide vehicle and the guide vehicle.

[0030] Furthermore, when the traversable node position is outside the area to be searched, the intersection of a line segment formed by the position of the guide vehicle and the traversable node position and a boundary of the area to be searched is set as the traversable node position.

[0031] Furthermore, the to-be-explored stack adopts a first-in, last-out strategy to store node positions.

[0032] Furthermore, when a target is found, based on the predicted motion trajectory, the three vehicles of the current group surround and capture the target in a triangular manner, including:

[0033] Constructing a first equilateral triangle by taking the current position of the target as the center of the first equilateral triangle and taking the predicted motion trajectory as the midline of the first equilateral triangle;

[0034] The second equilateral triangle is constructed by taking the predicted end position of the target at a time interval as the center of the second equilateral triangle and taking the continuously predicted motion trajectory as the midline of the second equilateral triangle;

[0035] The guide vehicle and the two member vehicles respectively use the vertices of the adjacent first equilateral triangle and the second equilateral triangle to construct a moving path based on a cubic polynomial spline curve and perform uniform sampling to obtain the control points of the guide vehicle and the two member vehicles to encircle and capture the target.

[0036] Furthermore, the guide vehicle predicts the motion trajectory of the target based on the HIVT algorithm.

[0037] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects:

[0038] 1. In the present invention, each group guide vehicle performs incremental environmental node exploration based on its own position and continuously conducts target exploration, which can quickly establish key nodes of the unknown environment and improve the exploration speed; the explored node positions are stored in a shared explored stack, and the exploration results are shared with each group guide vehicle, which reduces unnecessary duplication of work and improves the exploration efficiency.

[0039] 2. During the process of target capture, the present invention can effectively plan the capture path by accurately predicting the future position of the target, reducing the risk of losing the target. The cubic polynomial spline curve is used to generate a smooth tracking path, which not only reduces the impact of mechanical movement but also improves the response speed of the system.

[0040] 3. The present invention constructs a hierarchical target search and capture architecture, combining the advantages of centralized and distributed control. The unmanned vehicle cluster is divided into groups through the main control console, and the target area that needs to perform the search and capture mission is sent to the guide vehicle in the group. When the guide vehicle finds the target during the environmental exploration process, it will jointly capture the target with the member vehicles of the group. This not only ensures the overall coordination of the system but also improves the flexibility of local decision-making. It can adapt to a variety of different types of task requirements and improves the adaptability and reusability of the system.

[0041] In the present invention, the above-mentioned technical solutions can also be combined with each other to achieve more preferred combination solutions. Other features and advantages of the present invention will be described in the subsequent description, and some advantages can become obvious from the description, or can be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained through the contents particularly pointed out in the description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The drawings are only for the purpose of illustrating particular embodiments and are not to be considered limiting of the present invention. Like reference symbols denote like components throughout the drawings.

[0043] Figure 1 A schematic diagram of a process flow of a hierarchical target search and capture method for an unmanned vehicle cluster in an embodiment of the present invention;

[0044] Figure 2 It is a schematic diagram of the structure of a hierarchical target search and capture architecture in an embodiment of the present invention;

[0045] Figure 3 It is a schematic diagram of the process of triangular encirclement and capture in an embodiment of the present invention. DETAILED DESCRIPTION

[0046] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0047] A specific embodiment of the present invention discloses a hierarchical target search and capture method for unmanned vehicle clusters, such as Figure 1 As shown, the steps S1 to S4 are included:

[0048] Step S1: Divide the unmanned vehicle cluster into several groups, each of which includes a guide vehicle and two member vehicles.

[0049] Specifically, Figure 2 As shown, the present invention first constructs a hierarchical target search and capture architecture, wherein the nodes at the same level in the hierarchical architecture are distributed management structures, and the upper-level nodes use a centralized management architecture for the corresponding nodes at the lower level, including:

[0050] The first level is the main control console, which is used to divide the unmanned vehicle cluster into several groups and send the search area and target pictures to be searched for the search and capture mission to each group of guide vehicles for target exploration.

[0051] The second level is the guide vehicle of each group, which is used to conduct incremental exploration of environmental nodes starting from its own position and share the exploration map in real time until a group captures the escaped target position or fails to find the target in the target area. When the target is found, it builds its own moving path based on the target's motion trajectory to encircle the target.

[0052] The third level is for each team member vehicle, which is used to receive the adjacent vertex position information of the equilateral triangle generated by the guide vehicle according to the motion trajectory of the target to be captured when the guide vehicle of the group finds the target, and construct their own moving paths to capture the target together with the guide vehicle of the group.

[0053] Step S2: In the area to be searched, each group of guide vehicles searches for the target based on its own position.

[0054] Specifically, each group of guide vehicles uses its own sensors, such as lidar, camera, etc., to search for targets in the surrounding environment based on the target image to be searched sent by the main control console. The target search algorithm includes but is not limited to the R-CNN algorithm or the YOLO algorithm, which is not limited in the present invention.

[0055] It should be noted that when the target recognition probability value obtained by the target search algorithm is higher than a preset value, it is determined that the target is found. In this embodiment, this preset value is set to 0.85.

[0056] Step S3, when the target is not found, each group guide vehicle performs incremental environment node exploration based on its own position to continue the target search, including: storing each current position in the explored stack and moving to the latest node position in the corresponding to-be-explored stack to perform the target search until the search target is found or the corresponding node in the to-be-explored stack is empty; wherein, the to-be-explored stack is a storage space for the guide vehicle to perform radial node exploration at the target search position to obtain several passable node positions in the to-be-searched area; the explored stack is used to store the node positions that have been reached by the guide vehicles of each group.

[0057] Specifically, each group's to-be-explored stack stores the locations of traversable nodes obtained by the group's guide vehicle through radial node exploration, which is used to guide the search path. When the guide vehicle reaches the node location and performs target search, this node location is stored in the to-be-explored stack shared by each group, and the locations are shared between groups in real time to avoid repeated exploration of the same area by different groups.

[0058] Furthermore, the to-be-explored stack adopts a first-in, last-out strategy to store node positions.

[0059] Specifically, the first-in, last-out strategy enables the guide vehicle to conduct in-depth exploration in an unknown environment, which can reduce the guide vehicle's wandering around the explored area, thereby saving energy and time.

[0060] It should be noted that the explored stack uses an unordered container structure, which can be, for example, a hash table, a set or a dynamic array. Usually, containers of this structure can provide a fast search function, so as to quickly determine whether a node has been explored, thereby avoiding repeated exploration; and, since unordered containers do not need to maintain the order of elements, the overhead of sorting operations is avoided, which is very important for real-time systems.

[0061] Furthermore, the incremental environment node exploration stores the initial position of each group of guide vehicles into their respective to-be-explored stacks; the search process includes steps S301 to S305:

[0062] Step S301: Use the latest node position in the to-be-explored stack as the node position for this search and delete the node position from the to-be-explored stack.

[0063] Specifically, since the to-be-explored stack stores adjacent node positions using a first-in, last-out strategy, the latest node position is the node position that was most recently added to the stack.

[0064] Step S302: When the node position does not exist in the explored stack, the guide vehicle is moved to the node position for target search, and the node position is stored in the explored stack.

[0065] Specifically, the obtained node position does not exist in the explored stack, which means that this node position has not been explored by other group guide vehicles. At this time, the current guide vehicle moves to this node position for target retrieval.

[0066] If the obtained node position already exists in the explored stack, it means that it has been explored by other group guide vehicles. Then, it is necessary to obtain the latest node position in the to-be-explored stack of the current guide vehicle again and perform judgment search again.

[0067] Step S303: If the target is not found during the search, a 360° radial node exploration is performed with the guide vehicle as the center to obtain the positions of several traversable nodes in the search area.

[0068] It should be noted that when the target recognition probability value obtained by the target search algorithm is lower than a preset value, it is determined that the target is not found. In this embodiment, this preset value is set to 0.85.

[0069] Specifically, 360° radial node exploration enables the guide vehicle to fully detect the surrounding environment, discover and record all potential traversable nodes, reduce search blind spots, and improve search efficiency and coverage.

[0070] Further, obtaining the positions of several traversable nodes in the area to be searched includes steps S3031 to S3034:

[0071] Step S3031, with the guide vehicle as the center, the front of the vehicle body as 0°, and preset angles as intervals, the environment is continuously scanned 360° counterclockwise to obtain the distance between objects at various angles and the guide vehicle.

[0072] Specifically, using sensors carried by the guide vehicle, such as a lidar sensor or an infrared sensor, with the position of the guide vehicle as the center, the front of the vehicle is facing straight ahead at 0°, and rays are emitted at intervals at preset angles to obtain the distance between objects and the guide vehicle at various angles; illustratively, the preset angle is determined based on target detection requirements, environmental complexity or sensor type. In this embodiment, it can be set to 5°.

[0073] Furthermore, when the distance between the object and the guide vehicle at each angle is obtained by continuous scanning, when the distance is greater than a preset maximum distance, the preset maximum distance is used as the distance between the object and the guide vehicle at the current angle.

[0074] Specifically, when the sensor carried by the guide vehicle transmits back an object that is too far away or has not explored the object at a certain angle, in order to ensure that the guide vehicle has a certain coverage range of the surrounding environment when searching for the target and avoid missing the target during the search process, based on the accuracy of the sensor carried by the guide vehicle and the complexity of the surrounding environment, a maximum distance is preset as the distance between the object at the current angle and the guide vehicle. By way of example, in this embodiment, the preset maximum distance is 1000 meters.

[0075] Step S3032: construct an angle distance array based on the angle and the object distance obtained by scanning the angle.

[0076] Specifically, the angle distance array is used to record the position and distance of objects around the guide vehicle. Exemplarily, the angle distance array can be expressed as {(θ1, d1), (θ2, d2), ..., (θ n ,d n )}, where θ n represents the nth scanning angle, d n Represents the distance between the object and the guided vehicle at the nth scanning angle.

[0077] Step S3033, based on the angle distance array, obtaining the position of each object around the guide vehicle, comprising:

[0078] Traversing the angle distance array and reading the data of the angle distance array;

[0079] Based on the position of the guide vehicle, each angle and distance in the angle distance array, the position of each object around the guide vehicle is obtained; wherein, when the difference between the distances of consecutive angles in the angle distance array is within the distance fluctuation threshold, the objects in this continuous angle range are regarded as continuous objects, the midpoint of this continuous angle range is taken as the angle of the object relative to the guide vehicle, and the shortest distance in the continuous angle range is taken as the distance to the guide vehicle.

[0080] Specifically, by identifying continuous objects in the environment, more accurate environmental information can be provided to the guide vehicle, reducing the number of node locations that need to be reached and improving the efficiency of target search.

[0081] Exemplarily, the preset distance fluctuation threshold is 10 meters. When the angles are 10°, 15°, and 20°, the measured distances are 500 meters, 505 meters, and 497 meters respectively, and the distance differences are 5 meters and 7 meters respectively. At this time, the objects measured at the three angles are regarded as a continuous object, and the angle of this continuous object relative to the guide vehicle is set to 10°+[(15°-10°)+(20°-15°)] / 2=15°. In order to prevent the guide vehicle from colliding with this object when it arrives at the node position, the distance of this continuous object relative to the guide vehicle is set to the shortest distance of 497 meters.

[0082] It should be noted that when the distance between an object and the guide vehicle is within a preset distance, the object is judged to be a close-range obstacle and not a passable node, and it needs to be deleted from the angle distance array. For example, in this embodiment, the preset distance is set to 100 meters.

[0083] Furthermore, based on the position of the guide vehicle, each angle and distance in the angle distance array, the position of each object around the guide vehicle is obtained using the following formula:

[0084]

[0085] Among them, (x i ,y i ) represents the position coordinates of the i-th object around the guide vehicle; (x guide ,y guide ) represents the position coordinates of the guide vehicle; θ i represents the angle of the i-th object around the guide vehicle relative to the guide vehicle; d i Represents the distance between the i-th object around the guide vehicle and the guide vehicle.

[0086] Step S3034: subtract the preset safety distance from the position of each object to obtain the corresponding passable node position.

[0087] Specifically, in order to ensure that the guide vehicle maintains a certain safe distance from obstacles during driving and reduce the risk of collision, a retraction safety distance is set on the straight-line distance from the guide vehicle to the position of each object, and this position is set as a passable node position.

[0088] Exemplarily, the coordinates of the traversable node positions are obtained using the following formula:

[0089]

[0090] Among them, (x node,i ,ynode,i ) represents the position coordinates of the ith passable node; s represents the safety distance.

[0091] Furthermore, when the traversable node position is outside the area to be searched, the intersection of a line segment formed by the position of the guide vehicle and the traversable node position and a boundary of the area to be searched is set as the traversable node position.

[0092] Specifically, when the calculated position of the traversable node is outside the area to be searched, based on the traversable node position coordinates, the guide vehicle position coordinates and the boundary of the area to be searched, the line segment and polygon intersection algorithm is used to calculate the intersection position coordinates of the line segment formed by the guide vehicle position and the traversable node position and the boundary of the area to be searched, and use them as the traversable node position coordinates.

[0093] Step S304: Store the traversable node positions that do not exist in the explored stack or in the to-be-explored stack in the to-be-explored stack.

[0094] Specifically, when the traversable node position is a position that has been visited by other group guide vehicles, or has been explored and stored in the current to-be-explored stack, this node position does not need to be explored again.

[0095] Step S305: perform the next search until the search target is found or the to-be-explored stack is empty, and then terminate the search.

[0096] Step S4: When a target is found, the guide vehicle of the group predicts the motion trajectory of the target; based on the predicted motion trajectory, the three vehicles of the current group surround and capture the target in a triangular manner.

[0097] Specifically, when the target recognition probability value obtained by the target search algorithm is higher than a preset value, it is determined that the target is found. In this embodiment, the preset value is set to 0.85.

[0098] Furthermore, the guide vehicle predicts the motion trajectory of the target based on the HIVT algorithm.

[0099] Specifically, the HIVT (Hybrid Intelligent Vehicle Tracking) algorithm is an algorithm for target tracking. It extracts features of the target based on the sensors carried by the guide vehicle and continuously tracks the target. It also collects the target's motion data, including the target's position, speed, acceleration and other information, and predicts its future motion trajectory based on the target's history and current state.

[0100] Furthermore, based on the predicted motion trajectory, the three vehicles of the current group perform a triangular encirclement and capture of the target, including steps S401 to S403:

[0101] Step S401: construct a first equilateral triangle with the current position of the target as the center of the first equilateral triangle and the predicted motion trajectory as the midline of the first equilateral triangle.

[0102] Specifically, the midline of the triangle is a line segment connecting a vertex and the midpoint of its opposite side. It takes the current position of the target as the starting point and extends in the reverse direction along the predicted motion trajectory. The equilateral triangle construction algorithm is used to obtain the three vertices of the first equilateral triangle.

[0103] Step S402: construct a second equilateral triangle by taking the predicted end position of the target at a time interval as the center of a second equilateral triangle and taking the continuously predicted motion trajectory as the midline of the second equilateral triangle.

[0104] Specifically, the predicted terminal position is based on the motion data of the target collected by the guide vehicle for continuous tracking of the target, including information such as position, speed, acceleration, etc., and is predicted using the HIVT algorithm.

[0105] By constructing the second equilateral triangle, a capture point based on the predicted future position is provided, which enables the swarm of unmanned vehicles to plan the movement path in advance so that they can effectively capture the target when it reaches the predicted position.

[0106] Step S403: The guide vehicle and the two member vehicles respectively use the vertices of the adjacent first equilateral triangle and the second equilateral triangle to construct a moving path based on a cubic polynomial spline curve and perform uniform sampling to obtain control points of the guide vehicle and the two member vehicles to encircle the target.

[0107] Specifically, in this embodiment, the motion trajectory prediction is performed based on the guide vehicle in the group facing the target. Therefore, the vertex in the opposite direction of the target motion trajectory is the vertex of the equilateral triangle closest to the guide vehicle, such as Figure 3 As shown, the first equilateral triangle is triangle α, the second equilateral triangle is triangle β, and the vertices adjacent to the guide vehicle are vertex a of triangle α and vertex b of triangle β respectively.

[0108] The cubic polynomial spline curve creates a smooth and continuous curve through given data points, that is, based on the position of the guide vehicle, the vertex a of triangle α and the vertex b of triangle β, by constructing a cubic polynomial between each pair of adjacent data points and ensuring that these polynomials not only have equal values ​​at the data points, but also have equal first-order and second-order derivatives, thereby achieving highly smooth transition and continuity on the entire curve.

[0109] The member vehicles and the adjacent vertices of the first equilateral triangle and the second equilateral triangle use the same cubic polynomial spline curve to construct a moving path, so that the entire team gradually forms a triangular encirclement of the escaping target and completes the capture.

[0110] In summary, the hierarchical target search and capture method for unmanned vehicle clusters according to the embodiment of the present invention has the following beneficial effects:

[0111] 1. In the present invention, each group guide vehicle performs incremental environmental node exploration based on its own position and continuously conducts target exploration, which can quickly establish key nodes of the unknown environment and improve the exploration speed; the explored node positions are stored in a shared explored stack, and the exploration results are shared with each group guide vehicle, which reduces unnecessary duplication of work and improves the exploration efficiency.

[0112] 2. During the process of target capture, the present invention can effectively plan the capture path by accurately predicting the future position of the target, reducing the risk of losing the target. The cubic polynomial spline curve is used to generate a smooth tracking path, which not only reduces the impact of mechanical movement but also improves the response speed of the system.

[0113] 3. The present invention constructs a hierarchical target search and capture architecture, combining the advantages of centralized and distributed control. The unmanned vehicle cluster is divided into groups through the main control console, and the target area that needs to perform the search and capture mission is sent to the guide vehicle in the group. When the guide vehicle finds the target during the environmental exploration process, it will jointly capture the target with the member vehicles of the group. This not only ensures the overall coordination of the system but also improves the flexibility of local decision-making. It can adapt to a variety of different types of task requirements and improves the adaptability and reusability of the system.

[0114] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A hierarchical target search and capture method for unmanned vehicle clusters, characterized in that: The steps include: Divide the unmanned vehicle cluster into several groups, each of which includes a leader vehicle and two member vehicles; In the area to be searched, each group of guide vehicles searches for targets based on its own position; When the target is not found, each group of guide vehicles respectively conducts incremental environment node exploration based on its own position to continue the target search, including: storing each current position in the explored stack and moving to the latest node position in the corresponding to-be-explored stack to search for the target until the search target is found or the node in the corresponding to-be-explored stack is empty; wherein, the to-be-explored stack is a storage space for the guide vehicle to perform radial node exploration at the target search position to obtain several passable node positions in the to-be-searched area; the explored stack is used to store the node positions that each group of guide vehicles has reached; When a target is found, the guide vehicle of the group predicts the motion trajectory of the target; based on the predicted motion trajectory, the three vehicles of the current group surround and capture the target in a triangular manner.

2. The method according to claim 1, characterized in that: The incremental environment node exploration stores the initial position of each group of guide vehicles into their respective to-be-explored stacks. During the search process: Using the latest node position in the to-be-explored stack as the node position for this search and deleting the node position from the to-be-explored stack; When the node position does not exist in the explored stack, moving the guide vehicle to the node position for target search, and storing the node position in the explored stack; If the search does not find the target, a 360° radial node exploration is performed with the guide vehicle as the center to obtain several traversable node positions in the area to be searched; and the traversable node positions that do not exist in the explored stack or in the to-be-explored stack are stored in the to-be-explored stack; The next search is performed until the search target is found or the to-be-explored stack is empty, and the search is terminated.

3. The method according to claim 2, characterized in that: Get the locations of several traversable nodes in the area to be searched, including: With the guide vehicle as the center and the front of the vehicle body as 0°, the environment is continuously scanned 360° counterclockwise at preset angles to obtain the distance between the object and the guide vehicle at each angle; Constructing an angle distance array based on the angle and the object distance obtained by scanning the angle; Based on the angle distance array, obtaining the position of each object around the guide vehicle; Subtract the preset safety distance from the position of each object to obtain the corresponding passable node position.

4. The method according to claim 3, characterized in that: When the distance between the object and the guide vehicle at each angle is obtained by continuous scanning, when the distance is greater than a preset maximum distance, the preset maximum distance is used as the distance between the object and the guide vehicle at the current angle.

5. The method according to claim 4, characterized in that: The obtaining the position of each object around the guide vehicle based on the angle distance array includes: Traversing the angle distance array and reading the data of the angle distance array; Based on the position of the guide vehicle, each angle and distance in the angle distance array, the position of each object around the guide vehicle is obtained; wherein, when the difference between the distances of consecutive angles in the angle distance array is within the distance fluctuation threshold, the objects in this continuous angle range are regarded as continuous objects, the midpoint of this continuous angle range is taken as the angle of the object relative to the guide vehicle, and the shortest distance in the continuous angle range is taken as the distance to the guide vehicle.

6. The method according to claim 5, characterized in that: Based on the position of the guide vehicle, each angle and distance in the angle distance array, the position of each object around the guide vehicle is obtained using the following formula: Among them, (x i ,y i ) represents the position coordinates of the i-th object around the guide vehicle; (x guide ,y guide ) represents the position coordinates of the guide vehicle; θ i represents the angle of the i-th object around the guide vehicle relative to the guide vehicle; d i Represents the distance between the i-th object around the guide vehicle and the guide vehicle.

7. The method according to claim 3, characterized in that: When the traversable node position is outside the area to be searched, the intersection of the line segment formed by the position of the guide vehicle and the traversable node position and the boundary of the area to be searched is set as the traversable node position.

8. The method according to any one of claims 1 to 7, characterized in that: The to-be-explored stack uses a first-in, last-out strategy to store node positions.

9. The method according to claim 1, characterized in that: When a target is found, based on the predicted motion trajectory, the three vehicles of the current team will surround the target in a triangular manner, including: Constructing a first equilateral triangle by taking the current position of the target as the center of the first equilateral triangle and taking the predicted motion trajectory as the midline of the first equilateral triangle; The second equilateral triangle is constructed by taking the predicted end position of the target at a time interval as the center of the second equilateral triangle and taking the continuously predicted motion trajectory as the midline of the second equilateral triangle; The guide vehicle and the two member vehicles respectively use the vertices of the adjacent first equilateral triangle and the second equilateral triangle to construct a moving path based on a cubic polynomial spline curve and perform uniform sampling to obtain the control points of the guide vehicle and the two member vehicles to encircle and capture the target.

10. The method according to claim 9, characterized in that: The guide vehicle predicts the motion trajectory of the target based on the HIVT algorithm.

Images