AUV cluster sensing communication integrated data collection and path planning method

Abstract

The invention relates to an AUV (Autonomous Underwater Vehicle) cluster sensing communication integrated data collection and path planning method, which is applied to an underwater sensing communication integrated system, the system comprises a data center, a plurality of underwater control nodes, an AUV cluster and a plurality of sensor nodes, and the method comprises the following steps: the data center allocates a working area for each underwater control node; each control node divides the sensor nodes in the working area of the control node into a plurality of sensor node clusters and calculates the overlapping communication area of each sensor node cluster, and then plans a path for the AUV cluster to which the control node belongs, so that the energy efficiency is maximized when the AUV cluster cooperatively traverses the overlapping communication area of each sensor node cluster in the working area to complete data collection. According to the invention, the energy efficiency of the AUV underwater wireless sensor network can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of perception and communication integration, and in particular to an AUV cluster perception and communication integrated data collection and path planning method. Background Art

[0002] With the development of ocean exploration, seabed data collection plays an important role in underwater operations. As the main mode of underwater remote communication, hydroacoustic network is an indispensable means of data collection. Autonomous underwater vehicle (AUV) is a maneuverable and controllable device, and its auxiliary data collection strategy can improve the performance of water'xia.

[0003] In underwater wireless sensor networks, sensor nodes are installed underwater, and their battery capacity limits their working time, thus limiting the life of the network. AUVs can sail to the vicinity of sensor nodes for data collection, which greatly saves the energy consumed by the sensor's data transmission, thereby extending the life of the network. However, in a dynamic ocean environment, AUV-assisted data collection also faces the problem of difficulty in online planning. Therefore, an effective AUV data collection and trajectory planning strategy is needed.

[0004] There is a wide range of research work on collaborative data collection and path planning strategies for AUV swarms. Some of this work focuses on designing fixed trajectories based on the location of sensor nodes and prior knowledge of the ocean environment, such as obstacles and currents. However, due to the dynamic changes in the ocean environment, it is unrealistic to know the location of obstacles in advance. In some other studies, the application of reinforcement learning has achieved online path planning for AUVs based on obstacle avoidance, but this also requires the use of sonar to detect obstacles in the environment in real time. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide an AUV cluster perception communication integrated data collection and path planning method, which can improve the energy efficiency of the AUV underwater wireless sensor network.

[0006] The technical solution adopted by the present invention to solve the technical problem is: to provide an AUV cluster perception communication integrated data collection and path planning method, which is applied to an underwater perception communication integrated system, the system includes a data center, a plurality of underwater control nodes and an AUV cluster, and a plurality of sensor nodes, including the following steps:

[0007] The data center allocates a working area for each underwater control node;

[0008] Each control node divides the sensor nodes in its working area into multiple sensor node clusters and calculates the overlapping communication area of ​​each sensor node cluster, and then plans a path for the AUV cluster to which it belongs, so that the energy efficiency of the AUV cluster can be maximized when collaboratively traversing the overlapping communication areas of each sensor node cluster in the working area to complete data collection.

[0009] Furthermore, the sensor nodes in the working area are divided into a plurality of sensor node clusters, including:

[0010] According to the number of overlapping communication areas between each sensor node and other sensor nodes in the working area, the sensor nodes are sorted from most to least to obtain a sensor node sequence;

[0011] Establish a sensor node cluster with the first sequence node of the sensor node sequence as the cluster node;

[0012] Traverse the sensor node sequence, if the current sequence node has overlapping communication areas with all nodes in the current sensor node cluster, remove it from the sensor node sequence and put it into the sensor node cluster;

[0013] It is determined whether the current sensor node sequence is empty. If it is not empty, the step of establishing a sensor node cluster with the first sequence node of the sensor node sequence as a cluster node is returned. Otherwise, the overlapping communication areas of each sensor node cluster are respectively calculated.

[0014] Furthermore, the overlapping communication area of ​​the sensor node cluster is calculated by the following method:

[0015] Get the circular communication range of each cluster node and find the intersection between each circular communication range;

[0016] Determine whether the midpoint of the arc between any two intersection points falls within the range of all circular communications. If so, the arc is considered to be the boundary of the overlapping communication area.

[0017] Get all the boundaries of the overlapping communication area and calculate the center point of the overlapping communication area.

[0018] Furthermore, planning a path for the AUV cluster to which it belongs, so that the energy efficiency of the AUV cluster when collaboratively traversing the overlapping communication areas of each sensor node cluster in the working area to complete data collection is maximized, includes:

[0019] With the goal of maximizing the energy efficiency of AUV, the order in which each AUV in the cluster traverses the sensor node cluster is determined;

[0020] Based on the traversal order of the AUV for the sensor node cluster, a SAC-based path planning model is constructed to solve the shortest data collection path for each AUV and avoid obstacles during navigation.

[0021] Furthermore, the traversal order of the AUV cluster to the sensor node cluster is determined with the goal of maximizing the energy efficiency of the AUV, including:

[0022] Model the path planning problem as a min-max multiple traveling salesman problem;

[0023] Solve the minimum-maximum multi-traveling salesman problem and obtain the traversal order that minimizes the product of the total path length and the longest path.

[0024] Furthermore, the minimum-maximum multi-traveling salesman problem is solved by using an ant colony algorithm, and the best path after each iteration is evaluated by minimizing the product of the total path length and the longest path.

[0025] Furthermore, the state space of the path planning model includes the relative direction of the target location of the AUV's current voyage and the relative direction of the obstacle closest to the AUV; the action space of the path planning model is the forward direction of the AUV; and the reward function of the path planning model includes rewards for avoiding obstacles and approaching the target node, as well as penalties for collision, disconnection and timeout.

[0026] Furthermore, the AUV is equipped with a perception and communication integration module, which communicates with the control node and the sensor node through the perception and communication integration signal, and detects obstacles in the surrounding environment at the same time.

[0027] Furthermore, the detecting of obstacles in the surrounding environment includes:

[0028] The AUV obtains the perception and communication integrated signal reflected by the obstacle, extracts the transmitting node ID and transmission time after unpacking, and then calculates the distance between the AUV and the obstacle at the current position and sends it to the control node;

[0029] The control node uses the TOA algorithm to solve the position coordinates of the current obstacle according to the distance between the AUV and the same obstacle at at least three different positions.

[0030] Furthermore, the data center allocates a working area to each control node according to the locations of all control nodes and sensor nodes.

[0031] Beneficial Effects

[0032] Due to the adoption of the above technical scheme, the present invention has the following advantages and positive effects compared with the prior art: the present invention combines the underwater sensing and communication integration technology with the underwater wireless sensor network, so that the AUV cluster can detect and avoid obstacles on the way of collecting data, and proposes a clustering method based on overlapping communication areas to improve the energy efficiency of the AUV cluster. At the same time, the reinforcement learning technology is used to plan the movement path of the AUV cluster online, which overcomes the low efficiency of the traditional AUV-based underwater wireless sensor network, and improves the data throughput, energy efficiency of the sensor network and the flexibility and safety of the AUV cluster when collecting data. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a schematic diagram of the structure of an integrated network of underwater acoustic communication and detection according to an embodiment of the present invention;

[0034] Figure 2 It is a time slot allocation diagram of the perception and communication scheme according to an embodiment of the present invention. DETAILED DESCRIPTION

[0035] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall within the scope limited by the appended claims of the application equally.

[0036] The embodiment of the present invention relates to an integrated network data collection scheme for underwater acoustic communication detection based on an autonomous underwater vehicle (AUV) cluster, which is applied to an underwater perception and communication integrated system. The system includes a data center, a plurality of underwater control nodes and an AUV cluster, and a plurality of sensor nodes, and includes the following steps:

[0037] The data center allocates a working area for each underwater control node;

[0038] Each control node divides the sensor nodes in its working area into multiple sensor node clusters and calculates the overlapping communication area of ​​each sensor node cluster, and then plans a path for the AUV cluster to which it belongs, so that the energy efficiency of the AUV cluster can be maximized when collaboratively traversing the overlapping communication areas of each sensor node cluster in the working area to complete data collection.

[0039] The present invention will be described in detail below in conjunction with the accompanying drawings:

[0040] like Figure 1As shown in the figure, the Integrated Sensing and Communication for Underwater Acoustic Networks (ISAC-UANs) network consists of a ship-based data center, multiple underwater control nodes, sensor nodes, and AUV clusters. The data center stores the locations of all sensor nodes and control nodes, assigns each control node its working area, and collects the data collected by the control node to maintain the global network status information. The control node stores the locations of all sensor nodes in its working area, clusters the sensor nodes based on the location and communication range of these sensor nodes, and determines the overlapping communication area of ​​each cluster. Then, it plans the traversal order of each cluster by the AUV cluster to which it belongs to complete data collection, and summarizes the data and detection information collected from the AUV. In addition, all action decisions of the AUV are completed by the control node. The control node plans the path for the AUV to collect sensor node data and avoid obstacles through the location, detection and other information uploaded by the AUV. The AUV is equipped with an ISAC module, which communicates with the control node and sensor nodes through the ISAC signal and detects obstacles in the surrounding environment. AUVs controlled by the same control node can collaborate to perform data collection tasks. The sensors are equipped with various environmental monitoring functions. These sensor nodes are randomly distributed in the seabed environment. Sensor nodes with overlapping communication areas can be clustered, so the AUV can communicate with all nodes in the cluster by hovering in the overlapping communication area (preferably the center point).

[0041] like Figure 2 As shown in the figure, in the communication scheme, the completion of the data collection task is divided into multiple time rounds, including three steps: First, the control node sends commands to the AUV based on the AUV's position, data collection status, and detected obstacle positions. Second, the AUV periodically transmits ISAC signals during navigation to request sensor nodes to upload data and detect the environment. Finally, the sensor nodes receive the ISAC signals and upload the data to the AUV.

[0042] The AUV communicates with the sensor via ISAC signals, including search packets sent by the AUV and packets uploaded by the sensor node to the AUV. In addition to searching for nearby sensor nodes, the search packets sent by the AUV also have an environmental detection function. The controller communicates with the AUV via hydroacoustic communication signals, including downlink packets sent by the controller to the AUV and uplink packets sent by the AUV to the controller.

[0043] The ISAC search data packets sent by the AUV may be reflected by obstacles. When the AUV receives these echo packets, it can demodulate the packet header to obtain the source node ID and transmission time of the packet. Based on this information, the AUV can determine the distance the packet has traveled, thereby calculating the distance between the AUV's current position and the obstacle and sending it to the control node. After the AUV repeats the above operations at more than three different locations, the control node can use the TOA algorithm to locate the obstacle based on this information.

[0044] The calculation formula is as follows:

[0045]

[0046] where t A ,t B and t C represent the echo time delay observed at positions A, B and C respectively; v s is the speed of sound underwater; (x a ,y a )、(x b ,y b )、(x c ,y c ) and (x o ,y o ) represent the coordinates of positions A, B, C and obstacle O respectively.

[0047] In the path planning problem, the combinatorial optimization problem of maximizing the energy efficiency of the AUV cluster is split into a pre-planning stage and an online planning stage. In the pre-planning stage, we first cluster the nodes according to the overlapping communication areas and determine the traversal sequence that makes the AUV cluster most energy efficient. In the online planning stage, the AUV traverses the overlapping communication areas of each sensor node cluster according to the planned traversal order, uses ISAC signals for environmental perception, and performs real-time trajectory planning based on the environment. Therefore, we decompose the problem into sensor node clustering problem, cluster traversal problem, and trajectory planning problem.

[0048] The first sub-problem is the sensor node clustering problem, which requires dividing the sensor nodes into clusters and determining the overlapping communication area of ​​each cluster. The optimization goal is to minimize the number of node clusters, which can be solved by graph theory. Considering sensor nodes as vertices, if the communication ranges of two sensor nodes overlap, there is an edge between them. Within a cluster, the communication ranges of nodes overlap each other to form a cluster. Therefore, the clustering problem requires dividing the vertices into the smallest and largest clusters. To solve this problem, we adopt a clustering algorithm based on a greedy strategy and overlapping communication area. First, the total set of all sensor nodes is sorted based on the number of overlapping communication areas with other sensor nodes. Starting from the first node in the set, a temporary set is established to represent the cluster. If a node has communication edges with all nodes in the cluster, it will be added to the cluster and removed from the total geometry. Repeat this process until the total geometry is empty. Then, the overlapping communication area of ​​each cluster is calculated. The communication range of each node in the cluster can be represented as a circle. First, the intersection points between each pair of circles in the cluster are calculated to find all intersection points. These intersection points divide the circle into multiple arcs. Then, the midpoint of each arc is checked to determine if it lies inside all other circles. If so, the arc is considered one of the boundaries of the intersection region. After evaluating all arcs, the overlapping communication regions of the nodes in the cluster are determined. Finally, the center point of the overlapping communication region is calculated.

[0049] The second sub-problem is to determine the traversal order of the sensor node cluster. The goal of this sub-problem is to maximize the energy efficiency of the AUV, that is, the ratio of network throughput to consumed energy. The energy consumed is proportional to the total path length of the AUV cluster; and the network throughput is the ratio of the collected data to the time consumed to collect all data, and is inversely proportional to the longest path length of the AUV cluster. Therefore, the optimization goal can be equivalent to minimizing the product of the total path length and the longest path length. It can be modeled as a minimum-maximum multiple traveling salesman problem (Min-Max Multiple Travelling Salesman Problem, Min-Max MTSP). To solve this problem, an ant colony algorithm based on minimum-maximum value is used to solve it. Unlike the typical ant colony algorithm for solving the typical MTSP problem, the criterion for evaluating the best path obtained in each generation is to minimize the product of the total path length and the longest path, rather than minimizing the total path length.

[0050] The third sub-problem is to plan the shortest path from the starting position to the target position of the AUV and detect and avoid obstacles during navigation. In order to solve this optimization problem, we proposed a path planning algorithm based on soft actor-critic (SAC). Its state space, action space, reward function and training process are as follows:

[0051] State space: In addition to collecting information from each sensor node, the AUV also needs to automatically avoid obstacles. The state space of the system mainly consists of two parts, namely the relative direction of the target point and the relative direction of the nearest obstacle.

[0052] The action space is the forward direction of the AUV.

[0053] Reward function: The reward consists of the reward for avoiding obstacles and approaching the target node. In addition, the penalty includes collision penalty, disconnection penalty and timeout penalty.

[0054] Training process: The system uses the SAC algorithm, in which each AUV corresponds to four neural networks: action network, target action network, evaluation network and target evaluation network. First, the parameters of the four neural networks are initialized. In each action of each round, the action network decides the action to be taken according to the current state. After the AUV performs the action, it updates its own state. The environment rewards the AUV based on its state and action, and stores the state, action, reward and next state as a set of data in the experience buffer. After each round of training, several sets of data are taken from the experience buffer, and back propagation is performed according to the loss function of each neural network to update the parameters of each neural network.

Claims

1. A method for AUV cluster perception communication integrated data collection and path planning, applied to an underwater perception communication integrated system, the system includes a data center, several underwater control nodes and AUV clusters, and multiple sensor nodes, characterized in that: The following steps are involved: The data center allocates a working area for each underwater control node; Each control node divides the sensor nodes in its working area into multiple sensor node clusters and calculates the overlapping communication area of ​​each sensor node cluster, and then plans a path for the AUV cluster to which it belongs, so that the energy efficiency of the AUV cluster can be maximized when collaboratively traversing the overlapping communication areas of each sensor node cluster in the working area to complete data collection.

2. The method according to claim 1, characterized in that The sensor nodes in the working area are divided into a plurality of sensor node clusters, including: According to the number of overlapping communication areas between each sensor node and other sensor nodes in the working area, the sensor nodes are sorted from most to least to obtain a sensor node sequence; Establish a sensor node cluster with the first sequence node of the sensor node sequence as the cluster node; Traverse the sensor node sequence, if the current sequence node has overlapping communication areas with all nodes in the current sensor node cluster, remove it from the sensor node sequence and put it into the sensor node cluster; It is determined whether the current sensor node sequence is empty. If it is not empty, the step of establishing a sensor node cluster with the first sequence node of the sensor node sequence as a cluster node is returned. Otherwise, the overlapping communication areas of each sensor node cluster are respectively calculated.

3. The method according to claim 2, characterized in that The overlapping communication area of ​​a sensor node cluster is calculated by the following method: Get the circular communication range of each cluster node and find the intersection between each circular communication range; Determine whether the midpoint of the arc between any two intersection points falls within the range of all circular communications. If so, the arc is considered to be the boundary of the overlapping communication area. Get all the boundaries of the overlapping communication area and calculate the center point of the overlapping communication area.

4. The method according to claim 3, characterized in that The method of planning a path for the AUV cluster to which it belongs, so that the energy efficiency of the AUV cluster when collaboratively traversing the overlapping communication areas of each sensor node cluster in the working area to complete data collection is maximized, includes: With the goal of maximizing the energy efficiency of AUV, the traversal order of each AUV in the cluster for the sensor node cluster is determined; based on the traversal order of AUV for the sensor node cluster, a SAC-based path planning model is constructed to solve the shortest data collection path for each AUV separately and avoid obstacles during navigation.

5. The method according to claim 4, characterized in that The method aims to maximize the energy efficiency of the AUV and determines the traversal order of the AUV cluster for the sensor node cluster, including: Model the path planning problem as a min-max multiple traveling salesman problem; Solve the minimum-maximum multi-traveling salesman problem and obtain the traversal order that minimizes the product of the total path length and the longest path.

6. The method according to claim 5, characterized in that The minimum-maximum multi-traveling salesman problem is solved by using an ant colony algorithm, and the best path after each iteration is evaluated by minimizing the product of the total path length and the longest path.

7. The method according to claim 4, characterized in that The state space of the path planning model includes the relative direction of the target location of the AUV's current voyage and the relative direction of the obstacle closest to the AUV; the action space of the path planning model is the forward direction of the AUV; the reward function of the path planning model includes rewards for avoiding obstacles and approaching the target node, as well as penalties for collision, disconnection and timeout.

8. The method according to claim 4, characterized in that The AUV is equipped with a perception and communication integration module, which communicates with the control node and the sensor node through the perception and communication integration signal, and detects obstacles in the surrounding environment at the same time.

9. The method according to claim 8, characterized in that The detecting of obstacles in the surrounding environment comprises: The AUV obtains the perception and communication integrated signal reflected by the obstacle, unpacks it and extracts the transmitting node ID and transmission time. Then the distance between the AUV and the obstacle is calculated and sent to the control node; The control node uses the TOA algorithm to solve the position coordinates of the current obstacle according to the distance between the AUV and the same obstacle at at least three different positions.

10. The method according to claim 1, characterized in that The data center allocates a working area to each control node according to the locations of all control nodes and sensor nodes.

Images