An unmanned aerial vehicle cluster autonomous decision-making method and system fusing beidou enhanced service and multi-source perception

Abstract

The application discloses a kind of unmanned aerial vehicle cluster autonomous decision-making method and system of fusion Beidou enhanced service and multi-source perception, method includes: each unmanned aerial vehicle receives satellite-based augmentation signal, and real-time solution is carried out to satellite-based augmentation signal using precise point positioning technology, obtain precise PVT information;Each unmanned aerial vehicle synchronously collects sensor information, and with precise PVT information is spatiotemporal alignment, while using information fusion model processing local perception feature and wide-area geographic information, generate global environment situation tensor;Global environment situation tensor is input into deep reinforcement learning strategy network, obtain task instruction set;Using Beidou global short message, the key instruction in task instruction set is consensus confirmation, obtain consensus after task instruction;Each unmanned aerial vehicle is based on consensus after task instruction and cooperates work, and the new perception data generated in execution process is fused with precise PVT information in real time, input into information fusion model again, complete autonomous decision-making closed loop.

Description

Technical Field

[0001] This invention belongs to the field of unmanned system swarm control and artificial intelligence technology, specifically relating to an autonomous decision-making method and system for unmanned aerial vehicle swarms that integrates BeiDou enhanced services and multi-source perception. Background Technology

[0002] With the development of drone swarm technology, its application in fields such as emergency rescue and environmental monitoring is becoming increasingly widespread. However, in remote areas, oceans, or complex mountainous terrains without terrestrial network coverage, the capabilities of swarm systems face severe challenges: First, positioning accuracy is insufficient; traditional positioning methods such as GPS have meter-level accuracy, which is insufficient to meet the needs of precise collaborative operations. Second, communication link interruptions prevent collaborative decision-making within the swarm and with the rear. Third, the lack of a unified, high-precision spatiotemporal reference for perception and decision-making leads to distorted information fusion and a decline in decision quality.

[0003] The satellite-based augmentation and global short message services provided by the BeiDou-3 global satellite navigation system offer a unique infrastructure for solving the aforementioned problems. However, the current application of BeiDou technology in the field of unmanned aerial vehicles (UAVs) is mostly limited to single positioning or communication functions, failing to deeply integrate it into the intelligent decision-making closed loop of unmanned systems and failing to fully realize its potential as a "spatiotemporal information hub".

[0004] Therefore, there is an urgent need for an innovative solution that can systematically integrate BeiDou's cutting-edge services with advanced artificial intelligence decision-making technology to unlock the advanced autonomous capabilities of drone swarms in extreme environments. Summary of the Invention

[0005] This invention aims to address the shortcomings of existing technologies and provides the following solutions: An autonomous decision-making method for unmanned aerial vehicle (UAV) swarms that integrates BeiDou augmentation services and multi-source sensing includes the following steps: Each drone in the drone swarm receives satellite-based augmentation signals and uses precise point positioning technology to perform real-time calculations on the satellite-based augmentation signals to obtain precise PVT information; Each drone in the drone swarm synchronously collects sensor information and aligns it with the precise PVT information in time and space. At the same time, the information fusion model is used to process local perception features and wide-area geographic information to generate a global environmental situation tensor with time and space confidence. The global environment situation tensor is input into a pre-trained deep reinforcement learning policy network to obtain the task instruction set; The key instructions in the task instruction set are confirmed through consensus using BeiDou global short message service to obtain the consensus-based task instructions. Each UAV performs collaborative operations based on the consensus-based mission instructions, and integrates the new perception data generated during the execution process with the precise PVT information in real time, and then inputs it back into the information fusion model to complete the autonomous decision-making closed loop.

[0006] Preferably, the satellite-based augmentation signal is: in, Indicates receiver r For satellite s In frequency i pseudorange observations on Indicates the geometric distance between the receiver and the satellite. c Represents the speed of light. Indicates receiver clock bias. Indicates satellite clock bias, Indicates tropospheric delay, Indicates ionospheric delay, Indicates the receiver is at frequency i On pseudo-range hardware delay, Indicates the satellite's frequency i On pseudo-range hardware delay, This represents pseudorange observation noise and multipath error. Indicates receiver r For satellite s In frequency i Carrier phase observations on Represents frequency i carrier wavelength, Indicates the ambiguity of the integer period. Indicates receiver r In frequency i Phase hardware delay on, Indicates satellite s In frequency i Phase hardware delay on, This represents carrier phase observation noise and multipath error.

[0007] Preferably, the information fusion model includes: in, The attention score represents the spatiotemporal reliability. Q This represents the query vector in the attention mechanism. This represents the learnable weight matrix. Indicates from the i The semantic vector of each information source This represents the dimension of the key vector. λ This represents the preset weighting coefficient. Indicates the first i The spatiotemporal reliability of each information source.

[0008] Preferably, the method for performing the consensus confirmation includes: The leader node or decision-making node will receive the task instructions awaiting consensus. Cmd The data is then standardized and serialized, followed by computation using a collision-resistant cryptographic hash function to generate a fixed-length digital fingerprint, resulting in a hash digest. H cmd ; The hash digest H cmd Unique task number Task ID The key spatiotemporal parameters obtained from BeiDou precise positioning are encapsulated into a lightweight consensus request data packet. Pkt req Subsequently, the leadership node transmits the consensus request data packet via the BeiDou-3 RDSS short message service. Pkt req Broadcast to all follower nodes in the cluster and the remote command center; Upon receiving the consensus request data packet, the follower node or the command center Pkt re Then, perform the local verification step and receive a confirmation message. Msgack ; The leadership node is within a preset time window Δ T Internally, it listens for and collects the confirmation messages from other nodes. Msgack When the number of valid confirmation messages received M If the preset conditions are met, a consensus is considered to have been reached.

[0009] Preferably, the local verification step includes: Get complete task instructions Cmd’ ; The complete task instruction is computed using a collision-resistant cryptographic hash function. Cmd’ hash value H Cmd’ ; Compare the received hash digests H cmd The hash value calculated locally H Cmd’ If and only if H cmd = H Cmd’ At that time, the judgment instructions are consistent and have not been tampered with; If the command is found to be consistent and has not been tampered with, a confirmation message is sent in response via BeiDou short message. Msgack The confirmation message Msgack Includes the unique task number Task ID and confirmation signal ACK .

[0010] The present invention also provides an autonomous decision-making system for unmanned aerial vehicle (UAV) swarms that integrates BeiDou enhanced services and multi-source perception. The system applies the above-mentioned method and includes: an information acquisition module, an information processing module, a task instruction generation module, a consensus confirmation module, and an execution feedback module. In the information acquisition module, each UAV in the UAV cluster receives satellite-based augmentation signals and uses precise point positioning technology to perform real-time calculations on the satellite-based augmentation signals to obtain precise PVT information. In the information processing module, each UAV in the UAV cluster synchronously collects sensor information and aligns it with the precise PVT information in time and space. At the same time, the information fusion model is used to process local perception features and wide-area geographic information to generate a global environmental situation tensor with time and space confidence. The task instruction generation module is used to input the global environment situation tensor into a pre-trained deep reinforcement learning policy network to obtain a task instruction set. The consensus confirmation module uses BeiDou global short message service to confirm the key instructions in the task instruction set to obtain the consensus-received task instructions. In the execution feedback module, each UAV performs collaborative operations based on the consensus-based task instructions, and integrates the new perception data generated during the execution process with the precise PVT information in real time, and then inputs it back into the information fusion model to complete the autonomous decision-making closed loop.

[0011] Preferably, the satellite-based augmentation signal is: in, Indicates receiver r For satellite s In frequency i pseudorange observations on Indicates the geometric distance between the receiver and the satellite. c Represents the speed of light. Indicates receiver clock bias. Indicates satellite clock bias, Indicates tropospheric delay, Indicates ionospheric delay, Indicates the receiver is at frequency i On pseudo-range hardware delay, Indicates the satellite's frequencyi On pseudo-range hardware delay, This represents pseudorange observation noise and multipath error. Indicates receiver r For satellite s In frequency i Carrier phase observations on Represents frequency i carrier wavelength, Indicates the ambiguity of the integer period. Indicates receiver r In frequency i Phase hardware delay on, Indicates satellite s In frequency i Phase hardware delay on, This represents carrier phase observation noise and multipath error.

[0012] Preferably, the information fusion model includes: in, The attention score represents the spatiotemporal reliability. Q This represents the query vector in the attention mechanism. This represents the learnable weight matrix. Indicates from the i The semantic vector of each information source This represents the dimension of the key vector. λ This represents the preset weighting coefficient. Indicates the first i The spatiotemporal reliability of each information source.

[0013] Preferably, the workflow of the consensus confirmation module includes: The leader node or decision-making node will receive the task instructions awaiting consensus. Cmd The data is then standardized and serialized, followed by computation using a collision-resistant cryptographic hash function to generate a fixed-length digital fingerprint, resulting in a hash digest. H cmd ; The hash digest H cmd Unique task number Task ID The key spatiotemporal parameters obtained from BeiDou precise positioning are encapsulated into a lightweight consensus request data packet. Pkt req Subsequently, the leadership node transmits the consensus request data packet via the BeiDou-3 RDSS short message service. Pkt req Broadcast to all follower nodes in the cluster and the remote command center; Upon receiving the consensus request data packet, the follower node or the command center Pkt re Then, perform the local verification step and receive a confirmation message. Msgack ; The leadership node is within a preset time window Δ T Internally, it listens for and collects the confirmation messages from other nodes. Msgack When the number of valid confirmation messages received M If the preset conditions are met, a consensus is considered to have been reached.

[0014] Preferably, in the consensus confirmation module, the local verification step includes: Get complete task instructions Cmd’ ; The complete task instruction is computed using a collision-resistant cryptographic hash function. Cmd’ hash value H Cmd’ ; Compare the received hash digests H cmd The hash value calculated locally H Cmd’ If and only if H cmd = H Cmd’ At that time, the judgment instructions are consistent and have not been tampered with; If the command is found to be consistent and has not been tampered with, a confirmation message is sent in response via BeiDou short message. Msgack The confirmation message Msgack Includes the unique task number Task ID and confirmation signal ACK .

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention, through deep integration of the BeiDou-3 system's satellite-based augmentation and global short message communication capabilities, enables UAV swarms to maintain centimeter-level high-precision positioning and reliable communication in any area without terrestrial network coverage, significantly improving the system's environmental adaptability. Based on a unified spatiotemporal reference, a multi-source perception fusion mechanism deeply couples positioning accuracy with perception data, ensuring that decision-making processes such as path planning and target recognition are based on precise spatiotemporal data, greatly improving the accuracy and reliability of decisions. The innovatively designed lightweight consensus protocol fully utilizes the characteristics of BeiDou's short message service, ensuring the unimpeded flow of the core command chain even in extreme communication environments, endowing the system with strong anti-interference and fault tolerance capabilities. This invention deeply integrates the unique services of my country's independent BeiDou system, forming a unique and technologically advanced system-level solution with significant application value in key areas such as emergency rescue. Attached Figure Description

[0016] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the method flow according to an embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] Example 1 In this embodiment, as Figure 1 As shown, an autonomous decision-making method for UAV swarms that integrates BeiDou enhanced services and multi-source perception includes the following steps: S1. Each UAV in the UAV cluster receives satellite-based augmentation signals and uses precise point positioning technology to perform real-time calculations on the satellite-based augmentation signals to obtain precise PVT information.

[0021] In this embodiment, each unit in the UAV swarm uses an airborne BeiDou-3 receiver to receive PPP-B2b and / or Galileo HAS satellite-based augmentation signals. Through precise point positioning technology, it calculates and obtains absolute coordinates at the centimeter to decimeter level in real time. This high-precision position, velocity, and time information is used as a unified spatiotemporal reference to provide accurate spatiotemporal labels for all sensing data and subsequent decision-making.

[0022] The satellite-based augmentation signal is: in, Indicates receiver r For satellite s In frequency i pseudorange observations on Indicates the geometric distance between the receiver and the satellite. c Represents the speed of light. Indicates receiver clock bias. Indicates satellite clock bias, Indicates tropospheric delay, Indicates ionospheric delay, Indicates the receiver is at frequency i On pseudo-range hardware delay, Indicates the satellite's frequency i On pseudo-range hardware delay, This represents pseudorange observation noise and multipath error. Indicates receiver r For satellite s In frequency i Carrier phase observations on Represents frequency i carrier wavelength, Indicates the ambiguity of the integer period. Indicates receiver r In frequency i Phase hardware delay on, Indicates satellite s In frequency i Phase hardware delay on, This represents carrier phase observation noise and multipath error. Parameter estimation is performed using Kalman filtering, and the receiver's precise coordinates are calculated in real time to obtain accurate PVT information, providing a unified and high-precision spatiotemporal reference for each step of perception and decision-making.

[0023] S2. Each UAV in the UAV swarm synchronously collects sensor information and aligns it with precise PVT information in time and space. At the same time, the information fusion model is used to process local perception features and wide-area geographic information to generate a global environmental situation tensor with time and space confidence.

[0024] In this embodiment, the information fusion model based on the spatiotemporal attention mechanism introduces a spatiotemporal weight factor into its mathematical expression on top of the original attention mechanism: Let the set of semantic feature vectors from K information sources be . The corresponding BeiDou positioning information (accuracy factor, signal-to-noise ratio) constitutes a spatiotemporal reliability vector. The information fusion model includes: calculating an attention score that incorporates spatiotemporal reliability. in, This represents the final calculated attention score, which integrates content relevance and spatiotemporal reliability. Q This represents the query vector in the attention mechanism; This represents a learnable weight matrix used to combine input features. Projected onto the "key" space; Indicates from the i The semantic vector of each information source; Indicates the dimension of the key vector; This represents the standard scaled dot product attention calculation, used to evaluate the query and the first... i The relevance of content from each information source; λ This represents the preset weighting coefficient, used to balance the content relevance score and the spatiotemporal reliability score; Indicates the first i The spatiotemporal reliability of an information source is a real number. The introduced spatiotemporal reliability compensation term aims to add extra attention scores to data sources with higher positioning accuracy and better signal quality. This design allows the model to prioritize trusted sensing data with higher positioning accuracy and better spatiotemporal quality during fusion, in addition to focusing on the relevance of information content.

[0025] S3. Input the global environment situation tensor into a pre-trained deep reinforcement learning policy network to obtain the task instruction set.

[0026] In this embodiment, the global environmental situation tensor is input into a pre-trained deep reinforcement learning policy network. Based on its understanding of the spatiotemporal situation, the policy network outputs one or more high-level, abstract task instruction sets, which include: task type, target region and its spatiotemporal constraints.

[0027] S4. Use BeiDou global short message service to confirm the consensus on key instructions in the mission instruction set and obtain the consensus-based mission instructions.

[0028] In this embodiment, the method for consensus confirmation includes: The leader node or decision-making node will receive the task instructions awaiting consensus. CmdThe process involves standardizing and serializing the data, followed by computation using a collision-resistant cryptographic hash function (such as SHA-256) to generate a fixed-length digital fingerprint, thus obtaining the hash digest. H cmd : Here, Hash represents a cryptographic hash function. (This is a summary.) H cmd As a task instruction Cmd It is the only and unalterable representative.

[0029] hash digest H cmd Unique task number Task ID And key spatiotemporal parameters obtained from BeiDou precise positioning (such as the reference time for the command to take effect). T valid and center point coordinates P center Encapsulate it into a lightweight consensus request data packet. Pkt req : , Subsequently, the leader node transmits the consensus request data packet via the BeiDou-3 RDSS short message service. Pkt req Broadcast to all follower nodes in the cluster and the remote command center.

[0030] Follower nodes or command centers receive consensus request data packets Pkt re Then, perform the local verification step and receive a confirmation message. Msgack The local verification steps include: (1) obtaining complete task instructions. Cmd’ Specifically, complete mission instructions are received through a separate, higher-bandwidth communication link (such as a line-of-sight data link or Tiantong satellite broadband). Cmd’ or unique task number Task ID and center point coordinates P center Derive the expected task instructions from the pre-built task strategy library. Cmd’ (2) Use a collision-resistant cryptographic hash function to compute the complete task instructions. Cmd’ hash value H Cmd’ : (3) Compare the received hash digests H cmd Hash value calculated locally H Cmd’ If and only ifH cmd = H Cmd’ If the command is consistent and has not been tampered with, then a confirmation message is sent via BeiDou short message. Msgack Confirmation message Msgack It is a message sent via BeiDou short message service that contains at least a unique mission number. Task ID and confirmation signal ACK The structured acknowledgment message is a type of message where ACK is a positive response signal indicating "verification passed, acknowledgement received".

[0031] The leader node in the preset time window Δ T Internally, it listens for and collects confirmation messages from other nodes. Msgack When the number of valid confirmation messages received M If the preset conditions are met, a consensus is considered to have been reached. The preset conditions are: in, N This indicates the total number of nodes in the cluster, with the leader node itself counted as 1 acknowledgment. This indicates rounding up. This condition is equivalent to implementing most principles of distributed systems, allowing for rounding down to the nearest integer. The cluster detects node failures or malicious behavior. Once consensus is reached, the leader node broadcasts a task activation command, and all cluster members synchronously begin executing the task. Cmd .

[0032] S5. Each UAV performs collaborative operations based on the consensus-based mission instructions, and integrates the new perception data generated during the execution process with the precise PVT information in real time, and then inputs it back into the information fusion model to complete the autonomous decision-making closed loop.

[0033] Example 2 In this embodiment, an autonomous decision-making system for UAV swarms that integrates BeiDou enhanced services and multi-source perception includes: an information acquisition module, an information processing module, a task instruction generation module, a consensus confirmation module, and an execution feedback module.

[0034] In the information acquisition module, each UAV in the UAV cluster receives satellite-based augmentation signals and uses precise point positioning technology to perform real-time calculations on the satellite-based augmentation signals to obtain precise PVT information.

[0035] The satellite-based augmentation signal is: in, Indicates receiver r For satellite sIn frequency i pseudorange observations on Indicates the geometric distance between the receiver and the satellite. c Represents the speed of light. Indicates receiver clock bias. Indicates satellite clock bias, Indicates tropospheric delay, Indicates ionospheric delay, Indicates the receiver is at frequency i On pseudo-range hardware delay, Indicates the satellite's frequency i On pseudo-range hardware delay, This represents pseudorange observation noise and multipath error. Indicates receiver r For satellite s In frequency i Carrier phase observations on Represents frequency i carrier wavelength, Indicates the ambiguity of the integer period. Indicates receiver r In frequency i Phase hardware delay on, Indicates satellite s In frequency i Phase hardware delay on, This represents carrier phase observation noise and multipath error.

[0036] In the information processing module, each UAV in the UAV cluster synchronously collects sensor information and aligns it with the precise PVT information in time and space. At the same time, the information fusion model is used to process local perception features and wide-area geographic information to generate a global environmental situation tensor with spatiotemporal confidence.

[0037] Information fusion models include: in, The attention score represents the spatiotemporal reliability. Q This represents the query vector in the attention mechanism. This represents the learnable weight matrix. Indicates from the i The semantic vector of each information source This represents the dimension of the key vector. λ This represents the preset weighting coefficient. Indicates the first i The spatiotemporal reliability of each information source.

[0038] The task instruction generation module is used to input the global environment situation tensor into a pre-trained deep reinforcement learning policy network to obtain the task instruction set.

[0039] The consensus confirmation module uses BeiDou global short message service to confirm the key instructions in the task instruction set and obtain the consensus-based task instructions.

[0040] The consensus confirmation module's workflow includes: the leader node or decision node submits the task instructions to be agreed upon. Cmd The data is then standardized and serialized, followed by computation using a collision-resistant cryptographic hash function to generate a fixed-length digital fingerprint, resulting in a hash digest. H cmd ; hash digest H cmd Unique task number Task ID The key spatiotemporal parameters obtained from BeiDou precise positioning are encapsulated into a lightweight consensus request data packet. Pkt req Subsequently, the leading node transmits the consensus request data packet via the BeiDou-3 RDSS short message service. Pkt req Broadcast to all follower nodes in the cluster and the remote command center; follower nodes or the command center, upon receiving the consensus request data packet... Pkt re Then, perform the local verification step and receive a confirmation message. Msgack The leadership node is within the preset time window Δ T Internally, it listens for and collects confirmation messages from other nodes. Msgack When the number of valid confirmation messages received M If the preset conditions are met, a consensus is considered to have been reached.

[0041] In the consensus confirmation module, the local verification steps include: obtaining the complete task instructions. Cmd’ Use a collision-resistant cryptographic hash function to compute the complete task instructions. Cmd’ hash value H Cmd’ Compare the received hash digests. H cmd Hash value calculated locally H Cmd’ If and only if H cmd = H Cmd’ If the command is found to be consistent and unaltered, a confirmation message is sent via BeiDou short message service. Msgack Confirmation message Msgack Includes a unique task numberTask ID and confirmation signal ACK .

[0042] In the execution feedback module, each UAV performs collaborative operations based on the consensus-based task instructions, and integrates the new perception data generated during the execution process with the precise PVT information in real time, and then inputs it into the information fusion model to complete the autonomous decision-making closed loop.

[0043] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for autonomous decision-making in UAV swarms that integrates BeiDou enhanced services and multi-source sensing, characterized in that, Includes the following steps: Each drone in the drone swarm receives satellite-based augmentation signals and uses precise point positioning technology to perform real-time calculations on the satellite-based augmentation signals to obtain precise PVT information; Each drone in the drone swarm synchronously collects sensor information and aligns it with the precise PVT information in time and space. At the same time, the information fusion model is used to process local perception features and wide-area geographic information to generate a global environmental situation tensor with time and space confidence. The global environment situation tensor is input into a pre-trained deep reinforcement learning policy network to obtain the task instruction set; The key instructions in the task instruction set are confirmed through consensus using BeiDou global short message service to obtain the consensus-based task instructions. Each UAV performs collaborative operations based on the consensus-based task instructions, and integrates the new perception data generated during the execution process with the precise PVT information in real time, and then inputs it back into the information fusion model to complete the autonomous decision-making closed loop; The methods for achieving the consensus confirmation include: The leader node or decision-making node will receive the task instructions awaiting consensus. Cmd The data is then standardized and serialized, followed by computation using a collision-resistant cryptographic hash function to generate a fixed-length digital fingerprint, resulting in a hash digest. Hcmd ; The hash digest Hcmd Unique task number TaskID The key spatiotemporal parameters obtained from BeiDou precise positioning are encapsulated into a lightweight consensus request data packet. Pktreq Subsequently, the leadership node transmits the consensus request data packet via the BeiDou-3 RDSS short message service. Pktreq Broadcast to all follower nodes in the cluster and the remote command center; Upon receiving the consensus request data packet, the follower node or the command center Pktre Then, perform the local verification step and receive a confirmation message. Msgack ; The leadership node is within a preset time window Δ T Internally, it listens for and collects the confirmation messages from other nodes. Msgack When the number of valid confirmation messages received M If the preset conditions are met, a consensus is considered to have been reached.

2. The autonomous decision-making method for UAV swarms integrating BeiDou enhanced services and multi-source perception as described in claim 1, characterized in that, The satellite-based augmentation signal is: in, Indicates receiver r For satellite s In frequency i pseudorange observations on Indicates the geometric distance between the receiver and the satellite. c Represents the speed of light. Indicates receiver clock bias. Indicates satellite clock bias, Indicates tropospheric delay, Indicates ionospheric delay, Indicates the receiver is at frequency i On pseudo-range hardware delay, Indicates the satellite's frequency i On pseudo-range hardware delay, This represents pseudorange observation noise and multipath error. Indicates receiver r For satellite s In frequency i Carrier phase observations on Represents frequency i carrier wavelength, Indicates the ambiguity of the integer period. Indicates receiver r In frequency i Phase hardware delay on, Indicates satellite s In frequency i Phase hardware delay on, This represents carrier phase observation noise and multipath error.

3. The autonomous decision-making method for UAV swarms integrating BeiDou enhanced services and multi-source perception as described in claim 1, characterized in that, The information fusion model includes: in, The spatiotemporal reliability attention score represents the score. Q This represents the query vector in the attention mechanism. This represents the learnable weight matrix. Indicates from the i The semantic vector of each information source This represents the dimension of the key vector. λ This represents the preset weighting coefficient. Indicates the first i The spatiotemporal reliability of each information source.

4. The autonomous decision-making method for UAV swarms integrating BeiDou enhanced services and multi-source perception as described in claim 1, characterized in that, The local verification step includes: Get complete task instructions Cmd' ; The complete task instruction is computed using a collision-resistant cryptographic hash function. Cmd' hash value HCmd' ; Compare the received hash digests Hcmd The hash value calculated locally HCmd' If and only if Hcmd = HCmd' At that time, the judgment instructions are consistent and have not been tampered with; If the command is found to be consistent and has not been tampered with, a confirmation message is sent in response via BeiDou short message. Msgack The confirmation message Msgack Includes the unique task number TaskID and confirmation signal ACK .

5. An autonomous decision-making system for unmanned aerial vehicle (UAV) swarms integrating BeiDou enhanced services and multi-source perception, wherein the system applies the method described in any one of claims 1-4, characterized in that, include: The module includes an information acquisition module, an information processing module, a task instruction generation module, a consensus confirmation module, and an execution feedback module. In the information acquisition module, each UAV in the UAV cluster receives satellite-based augmentation signals and uses precise point positioning technology to perform real-time calculations on the satellite-based augmentation signals to obtain precise PVT information. In the information processing module, each UAV in the UAV cluster synchronously collects sensor information and aligns it with the precise PVT information in time and space. At the same time, the information fusion model is used to process local perception features and wide-area geographic information to generate a global environmental situation tensor with time and space confidence. The task instruction generation module is used to input the global environment situation tensor into a pre-trained deep reinforcement learning policy network to obtain a task instruction set. The consensus confirmation module uses BeiDou global short message service to confirm the key instructions in the task instruction set to obtain the consensus-received task instructions. In the execution feedback module, each UAV performs collaborative operations based on the consensus-based task instructions, and integrates the new perception data generated during the execution process with the precise PVT information in real time, and then inputs it back into the information fusion model to complete the autonomous decision-making closed loop.

6. The autonomous decision-making system for UAV swarms integrating BeiDou enhanced services and multi-source perception as described in claim 5, characterized in that, The satellite-based augmentation signal is: in, Indicates receiver r For satellite s In frequency i pseudorange observations on Indicates the geometric distance between the receiver and the satellite. c Represents the speed of light. Indicates receiver clock bias. Indicates satellite clock bias, Indicates tropospheric delay, Indicates ionospheric delay, Indicates the receiver is at frequency i On pseudo-range hardware delay, Indicates the satellite's frequency i On pseudo-range hardware delay, This represents pseudorange observation noise and multipath error. Indicates receiver r For satellite s In frequency i Carrier phase observations on Represents frequency i carrier wavelength, Indicates the ambiguity of the integer period. Indicates receiver r In frequency i Phase hardware delay on, Indicates satellite s In frequency i Phase hardware delay on, This represents carrier phase observation noise and multipath error.

7. The UAV swarm autonomous decision-making system integrating BeiDou enhanced services and multi-source perception as described in claim 5, characterized in that, The information fusion model includes: in, The spatiotemporal reliability attention score represents the score. Q This represents the query vector in the attention mechanism. This represents the learnable weight matrix. Indicates from the i The semantic vector of each information source This represents the dimension of the key vector. λ This represents the preset weighting coefficient. Indicates the first i The spatiotemporal reliability of each information source.

8. The autonomous decision-making system for UAV swarms integrating BeiDou enhanced services and multi-source perception as described in claim 5, characterized in that, The workflow of the consensus confirmation module includes: The leader node or decision-making node will receive the task instructions awaiting consensus. Cmd The data is then standardized and serialized, followed by computation using a collision-resistant cryptographic hash function to generate a fixed-length digital fingerprint, resulting in a hash digest. Hcmd ; The hash digest Hcmd Unique task number TaskID The key spatiotemporal parameters obtained from BeiDou precise positioning are encapsulated into a lightweight consensus request data packet. Pktreq Subsequently, the leadership node transmits the consensus request data packet via the BeiDou-3 RDSS short message service. Pktreq Broadcast to all follower nodes in the cluster and the remote command center; Upon receiving the consensus request data packet, the follower node or the command center Pktre Then, perform the local verification step and receive a confirmation message. Msgack ; The leadership node is within a preset time window Δ T Internally, it listens for and collects the confirmation messages from other nodes. Msgack When the number of valid confirmation messages received M If the preset conditions are met, a consensus is considered to have been reached.

9. The autonomous decision-making system for unmanned aerial vehicle (UAV) swarms integrating BeiDou enhanced services and multi-source perception as described in claim 8, characterized in that, In the consensus confirmation module, the local verification step includes: Get complete task instructions Cmd' ; The complete task instruction is computed using a collision-resistant cryptographic hash function. Cmd' hash value HCmd' ; Compare the received hash digests Hcmd The hash value calculated locally HCmd' If and only if Hcmd = HCmd' At that time, the judgment instructions are consistent and have not been tampered with; If the command is found to be consistent and has not been tampered with, a confirmation message is sent in response via BeiDou short message. Msgack The confirmation message Msgack Includes the unique task number Task ID and confirmation signal ACK .

Images