Unmanned system distributed data transmission method based on fuzzy confidence and hysteresis triggering

Abstract

The invention relates to an unmanned system distributed data transmission method based on fuzzy confidence and hysteresis triggering, and the method comprises the steps: carrying out the parallel extraction of an uncertainty feature vector and a task value feature vector corresponding to an unmanned system through target observation information collected by any unmanned system at a current moment; the uncertainty feature vector and the task value feature vector are subjected to double-layer fuzzy reasoning, and the comprehensive confidence degree corresponding to the unmanned system at the current moment is output; calculating a high trigger threshold and a low silence threshold based on the comprehensive confidence of the unmanned system at the previous moment; updating the data transmission state of the unmanned system at the current moment according to the hysteresis comparison result of the comprehensive confidence corresponding to the unmanned system at the current moment and the high trigger threshold value or the low silence threshold value; and when the updated data transmission state is active, the unmanned system executes a differentiated data transmission strategy according to the corresponding comprehensive confidence, so that reliable and real-time transmission of high-value and low-quality data in a low-bandwidth dynamic environment is realized.

Description

Technical Field

[0001] This application relates to the field of communication technology for unmanned systems, and in particular to a distributed data transmission method for unmanned systems based on fuzzy belief and hysteresis triggering. Background Technology

[0002] Existing unmanned systems use event-triggered distributed data transmission methods that reduce transmission by setting thresholds, but they still have the following prominent unresolved drawbacks: 1. Data confidence assessment and task value assessment are disconnected and rigid: Existing technologies usually calculate the confidence of data and task value independently, and then perform simple linear weighting or sequential judgment. This leads to the system delaying or discarding key information due to rigid quality judgment standards in the typical conflict scenario of "high task value and low observation confidence", and is unable to dynamically adjust the transmission decision according to the actual needs of the task.

[0003] 2. The triggering mechanism has poor anti-interference ability and is prone to network oscillation: Current methods mostly use a single static threshold triggering method. When there are slight fluctuations in data confidence or network status, frequent "trigger-silence" state switching, i.e. "ping-pong effect", is easily generated near the threshold, generating a large number of invalid control commands and aggravating network instability.

[0004] 3. Inefficient data transmission strategy and low bandwidth utilization: After triggering, a uniform compression encoding method is usually adopted, and the resource allocation is not refined according to the final value and quality assessment results of the data. Under extreme bandwidth-limited conditions, it may lead to link congestion or loss of high-value information. Summary of the Invention

[0005] To address the aforementioned issues, it is necessary to provide a distributed data transmission method for unmanned systems based on fuzzy belief and hysteresis triggering, including: S1: For any target observation information collected by an unmanned system at the current moment, extract the uncertainty feature vector and task value feature vector corresponding to the unmanned system in parallel; and output the comprehensive confidence level of the unmanned system at the current moment by passing the uncertainty feature vector and task value feature vector through two-layer fuzzy inference. S2: Calculate the high trigger threshold and low silence threshold based on the comprehensive confidence of the unmanned system at past times; update the data transmission status of the unmanned system at the current time based on the hysteresis comparison result between the comprehensive confidence of the unmanned system at the current time and the high trigger threshold or low silence threshold. S3: When the updated data transmission status is active, the unmanned system executes a differentiated data transmission strategy based on the corresponding comprehensive confidence level.

[0006] Preferably, the target observation information includes images, point clouds, or radio frequency signals.

[0007] Preferably, the uncertainty feature vector of the unmanned system is extracted, including: The quality index is obtained by calculating the structural similarity index of the target observation information; The occlusion ratio of the target observation information is calculated using a segmentation algorithm; Estimating motion ambiguity of unmanned systems based on inertial measurement unit data; Calculate the signal-to-noise ratio of the target observation information.

[0008] Preferably, the extracted task value feature vector of the unmanned system includes: The unmanned system acquires the task and calculates the normalized distance between the unmanned system and the target point in the task. Query the urgency of the retrieved tasks; Calculate the time-dependent decay factor based on the discovery duration.

[0009] Preferably, the uncertainty feature vector and the task value feature vector are subjected to two-layer fuzzy reasoning, including: Each element in the task value feature vector is fuzzified into its corresponding linguistic value through a membership function, and each linguistic value corresponding to the task value feature vector is mapped into a rule modulation factor according to a preset mapping table. Based on the rule modulation factor and the preset fuzzy mapping rule, the uncertainty feature vector is mapped to the comprehensive confidence level of the unmanned system at the current moment.

[0010] Preferably, the preset fuzzy mapping rule is: If the quality index is a first preset value, the occlusion ratio is a second preset value, the motion blur is a third preset value, and the signal-to-noise ratio is a fourth preset value, then the value calculated by the regular modulation factor will be used as the overall confidence level.

[0011] Preferably, the high trigger threshold and low silence threshold are calculated based on the comprehensive confidence level of the unmanned system at past times, including: A confidence sliding window of length W is set, and the confidence sliding window slides and samples the comprehensive confidence of the unmanned system at consecutive past moments with a preset step size. Calculate the mean and standard deviation of each composite confidence level in the confidence level sliding window corresponding to the current time. The high trigger threshold and low silence threshold are calculated based on the mean and the standard deviation, and the calculation formula includes: ; ; in, Indicates a high trigger threshold. Indicates a low silence threshold. This represents the mean of the overall confidence level. The standard deviation represents the overall confidence level. Indicates the first positive coefficient. Indicates the second positive coefficient. Indicates the third positive coefficient. This represents the network load factor at the current moment.

[0012] Preferably, the network load factor at the current moment includes the channel occupancy rate obtained by the unmanned system based on channel sensing at the current moment.

[0013] Preferably, updating the data transmission status of the unmanned system at the current moment includes: Obtain the data transmission status of the unmanned system at the current moment. If the comprehensive confidence level of the unmanned system at the current moment is greater than the high trigger threshold, and the data transmission status of the unmanned system at the current moment is silent, trigger data transmission and update the data transmission status of the unmanned system at the current moment to active. If the overall confidence level of the unmanned system at the current moment is less than the low silence threshold, and the data transmission status of the unmanned system at the current moment is active, then stop data transmission and update the data transmission status of the unmanned system at the current moment to silent. In all other cases, the data transmission status of the unmanned system remains unchanged at the current moment.

[0014] Preferably, S3 includes: When the updated data transmission status is active, if the overall confidence level of the unmanned system is greater than the first interval threshold, a lightweight video encoder is used to compress the target observation information collected by the unmanned system into the first data packet. If the overall confidence level of the unmanned system is greater than the second interval threshold and less than or equal to the first interval threshold, then the low-quality region in the target observation information is located according to the occlusion ratio and motion ambiguity. The super-resolution network is called to enhance the low-quality region in real time. The enhanced region is then stitched together with the other high-quality regions and compressed to obtain compressed data. A binary mask is generated to identify the enhanced region. The compressed data and the binary mask are then packaged into a second data packet. If the overall confidence level of the UAV system is less than or equal to the second interval threshold and the urgency of the task is greater than the urgency threshold, then the target category and bounding box in the target observation information are extracted through the lightweight target detection network, and the perceptual hash value of the target area is calculated. The timestamp, target category, bounding box, perceptual hash value and re-observation request flag are encapsulated into a third data packet. The first data packet, the second data packet, or the third data packet corresponding to the unmanned system is broadcast to the neighboring unmanned system through a distributed communication protocol.

[0015] Beneficial effects: This method extracts the uncertainty feature vector and task value feature vector corresponding to any unmanned system at the current moment by extracting target observation information collected by the unmanned system in parallel; it then performs two-layer fuzzy inference on the uncertainty feature vector and task value feature vector to output the comprehensive confidence level of the unmanned system at the current moment; based on the comprehensive confidence level of the unmanned system at past moments, it calculates the high trigger threshold and low silence threshold; based on the hysteresis comparison result between the comprehensive confidence level of the unmanned system at the current moment and the high trigger threshold or low silence threshold, it updates the data transmission status of the unmanned system at the current moment; when the updated data transmission status is active, the unmanned system executes a differentiated data transmission strategy according to the corresponding comprehensive confidence level, realizing reliable and real-time transmission of high-value, low-quality data in a low-bandwidth dynamic environment. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart of the distributed data transmission method for unmanned systems based on fuzzy belief and hysteresis triggering in the embodiments of this application. Detailed Implementation

[0018] To make the above-mentioned objectives, features, and advantages of this application more apparent and understandable, the specific embodiments of this application are described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of this application. However, this application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of this application. Therefore, this application is not limited to the specific embodiments disclosed below.

[0019] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0020] This embodiment provides a distributed data transmission method for unmanned systems based on fuzzy belief and hysteresis triggering. It is particularly suitable for distributed communication scenarios with low bandwidth, dynamic topology, and high task value perception requirements, such as disaster relief, environmental monitoring, or cluster collaboration tasks. This method is executed independently and in parallel by each unmanned system in the cluster. Figure 1 As shown, the method includes: S1: For any target observation information collected by an unmanned system at the current moment, extract the uncertainty feature vector and task value feature vector corresponding to the unmanned system in parallel; and output the comprehensive confidence level of the unmanned system at the current moment by passing the uncertainty feature vector and task value feature vector through two-layer fuzzy inference.

[0021] In this embodiment, the target observation information includes images, point clouds, or radio frequency signals.

[0022] Furthermore, the uncertainty feature vector of the unmanned system is extracted, including: The quality index is obtained by calculating the structural similarity index of the target observation information. The calculation formula is as follows: ; ; in, This represents the quality index corresponding to the i-th unmanned system at time t. Represents the structural similarity function. , These represent reference observation information and target observation information, respectively. , These represent the mean and variance of pixels in the reference observation information, respectively. , These represent the mean and variance of pixels in the target observation information, respectively. Describing covariance, , These represent the first stability constant and the second stability constant, respectively. The occlusion ratio of the target observation information is calculated using a segmentation algorithm, and the formula is as follows: ; ; ; ; ; in, This represents the occlusion ratio corresponding to the i-th unmanned system at time t, and its value range is... The higher the value, the greater the degree of occlusion. Indicates the area of ​​the visible region. Indicates the area of ​​the target region. Represents pixel coordinates, Indicates the target area identifier. Indicates visibility; The motion ambiguity of the unmanned system is estimated based on inertial measurement unit data. The formula for calculating the motion ambiguity is: ; ; in, This represents the motion ambiguity corresponding to the i-th unmanned system at time t, and its value range is... The larger the value, the more severe the blurring. Indicates the magnitude of angular velocity. Indicates the observation time. Represents the normalization constant. , , These represent the angular velocities of the horizontal, vertical, and longitudinal axes obtained by the inertial measurement unit, respectively. Calculate the signal-to-noise ratio of the target observation information.

[0023] Extracting the task value feature vector of the unmanned system, including: The unmanned system acquires the task, and the normalized distance between the unmanned system and the target point in the task is calculated. The formula for calculating the normalized distance is: ; ; in, This represents the normalized distance between the i-th unmanned system and the target point in the mission at time t, and its value ranges from 1 to 2. The smaller the value, the closer it is to the target. This represents the distance between the i-th unmanned system and the target point in the mission at time t. This represents the coordinates of the i-th unmanned system at time t. Represents the coordinates of the j-th task point; To query the urgency of the retrieved tasks, the formula for calculating task urgency is: ; ; in, This represents the task urgency of the i-th unmanned system at time t, and its value range is... A higher value indicates a more urgent task. Indicates the task type weight. Indicates the risk level. Indicates the urgency of time. , , These represent the first weighting coefficient, the second weighting coefficient, and the third weighting coefficient, respectively. The time-sensitivity decay factor based on the discovery duration is calculated using the following formula: ; in, This represents the time-degradation factor corresponding to the i-th unmanned system at time t. Indicates the time of the first observation. This represents the attenuation coefficient, and its value range is... The longer the time, the smaller the value.

[0024] Furthermore, the uncertainty feature vector and the task value feature vector are subjected to two-layer fuzzy reasoning, including: Each element in the task value feature vector is mapped to a corresponding rule modulation factor through a mapping function; Based on the rule modulation factor and the preset fuzzy mapping rule, the uncertainty feature vector is mapped to the comprehensive confidence level of the unmanned system at the current moment.

[0025] In this embodiment, the preset fuzzy mapping rule is: ; ; ; ; in, This represents the overall confidence level of the i-th unmanned system at time t. Let represent the contribution value of the k-th rule corresponding to the i-th unmanned system at time t. This represents the degree of matching of the k-th rule corresponding to the i-th unmanned system at time t. The larger the value, the more the rule matches the characteristics of the current observation information. When it is close to 1, it means that the rule is basically completely true, and when it is close to 0, it means that the rule is basically false. N represents the number of rules. This represents the rule modulation factor of the k-th rule corresponding to the i-th unmanned system at time t. This represents a mapping function, which can be a linear weighting function, a normalization function, or a nonlinear mapping function. It is used to map the task value feature vector to the regular modulation factor space. This represents the k-th element in the feature vector of the task value corresponding to the i-th unmanned system at time t. This represents the quality index of the i-th unmanned system corresponding to the k-th rule at time t. This represents the occlusion ratio of the i-th unmanned system corresponding to the k-th rule at time t. This represents the motion ambiguity of the i-th unmanned system corresponding to the k-th rule at time t. This represents the signal-to-noise ratio of the i-th unmanned system corresponding to the k-th rule at time t. , , , Let these represent the first, second, third, and fourth membership functions, respectively. Membership functions map continuous values ​​to membership degrees, thus characterizing the degree to which a certain indicator satisfies the corresponding semantic conditions. The membership degree values ​​range from [value missing]. .

[0026] The above fusion method uses a weighted average of the outputs of each rule, so that rules with higher activation levels have a greater influence in the final result, thereby ensuring that the evaluation results can reasonably reflect the overall state of the current observation information.

[0027] When the urgency of the task At extremely high levels, high A higher value will make the system more inclined to output a higher value. , The adjustment will weaken the rules regarding... and The stringent requirements mean that even with poor image quality, high output is still possible. .

[0028] S2: Calculate the high trigger threshold and low silence threshold based on the overall confidence level of the unmanned system at past times; update the data transmission status of the unmanned system at the current time based on the hysteresis comparison result between the overall confidence level of the unmanned system at the current time and the high trigger threshold or low silence threshold.

[0029] In this embodiment, the high trigger threshold and low silence threshold are calculated based on the comprehensive confidence level of the unmanned system at past times, including: A confidence sliding window of length W is set, and the confidence sliding window slides and samples the comprehensive confidence of the unmanned system at consecutive past moments with a preset step size. Calculate the mean and standard deviation of each composite confidence level in the confidence level sliding window corresponding to the current time. The high trigger threshold and low silence threshold are calculated based on the mean and the standard deviation, and the calculation formula includes: ; ; in, Indicates a high trigger threshold. Indicates a low silence threshold. This represents the mean of the overall confidence level. The standard deviation represents the overall confidence level. Indicates the first positive coefficient. Indicates the second positive coefficient. Indicates the third positive coefficient. This represents the network load factor at the current moment.

[0030] High trigger thresholds and low quiescent thresholds can be adaptively adjusted based on network load and historical comprehensive confidence statistics, thereby significantly suppressing frequent state switching caused by small fluctuations in evaluation values. This enables the triggering mechanism to have environmental and load awareness capabilities, thus preventing mass network congestion from the source.

[0031] In this embodiment, the network load factor at the current moment includes the channel occupancy rate obtained by the unmanned system based on channel sensing at the current moment.

[0032] Using the adaptive threshold calculation formula described above, firstly, the mean of the overall confidence score within the window is introduced. Standard deviation This allows the trigger threshold to dynamically fluctuate with the overall observation environment and task stage, avoiding over- or under-triggering issues caused by fixed thresholds when the environment changes, thus improving the system's robustness in different scenarios. Secondly, by introducing a network load factor... And adjust the high trigger threshold with its positive correlation. This mechanism automatically tightens the triggering conditions for data transmission when the network channel is busy, allowing only highly reliable critical information to be transmitted; conversely, it relaxes the triggering conditions when the network is idle. This achieves a distributed, locally-information-based congestion avoidance mechanism, optimizing the group's utilization efficiency of shared communication resources and fundamentally preventing network congestion and data collisions caused by multiple machines triggering simultaneously.

[0033] Furthermore, update the data transmission status of the unmanned system at the current moment, including: Obtain the data transmission status of the unmanned system at the current moment. If the comprehensive confidence level of the unmanned system at the current moment is greater than the high trigger threshold, and the data transmission status of the unmanned system at the current moment is silent, trigger data transmission and update the data transmission status of the unmanned system at the current moment to active. If the overall confidence level of the unmanned system at the current moment is less than the low silence threshold, and the data transmission status of the unmanned system at the current moment is active, then stop data transmission and update the data transmission status of the unmanned system at the current moment to silent. In all other cases, the data transmission status of the unmanned system remains unchanged at the current moment.

[0034] This hysteresis mechanism ensures that a change in transmission state is triggered only when the confidence level undergoes a sufficiently significant positive or negative jump, effectively filtering out channel fluctuations or minor noise in the data itself.

[0035] S3: When the updated data transmission status is active, the unmanned system executes a differentiated data transmission strategy based on the corresponding comprehensive confidence level.

[0036] Specifically, the steps include: When the updated data transmission status is active, if the overall confidence level of the unmanned system is greater than the first interval threshold, a high-confidence encoding process is executed: a lightweight video encoder is used to compress the target observation information collected by the unmanned system into a first data packet. ; If the overall confidence level of the unmanned system is greater than the second interval threshold and less than or equal to the first interval threshold, then the medium confidence enhancement coding process is executed: based on the occlusion ratio and motion ambiguity, low-quality regions in the target observation information are located; a super-resolution network is invoked to enhance the low-quality regions in real time; the enhanced regions are then concatenated with the remaining high-quality regions and compressed to obtain compressed data; a binary mask identifying the enhanced regions is generated; and the compressed data and the binary mask are packaged into a second data packet. ; If the overall confidence level of the UAV system is less than or equal to the second interval threshold and the mission urgency is greater than the urgency threshold, then the ground confidence metadata process is executed: The target category and bounding box are extracted from the target observation information using a lightweight target detection network, and the perceptual hash value of the target region is calculated. The timestamp, target category, bounding box, perceptual hash value, and re-observation request flag are encapsulated into a third data packet. ; If the overall confidence level of the UAV system is less than or equal to the second interval threshold and the mission urgency is less than or equal to the urgency threshold, the observation information is temporarily stored in a local cache queue, awaiting subsequent scheduling decisions; when cache resources are limited or exceed the preset storage time, this type of observation information can be discarded; wherein, the temporary storage time can be set as: ,in, Indicates the cache retention period. This indicates the maximum buffer time, thereby enabling adaptive management of low-value observation information, ensuring priority transmission of critical task information while improving the overall resource utilization efficiency of the system. The first data packet, the second data packet, or the third data packet corresponding to the unmanned system is broadcast to the neighboring unmanned system via a distributed communication protocol; and the neighboring unmanned system can determine the type of data received (e.g., (In the process of requesting repeated observations) to conduct collaborative sensing and information complementarity.

[0037] In this embodiment, the perceptual hash value of the target area is calculated, including: To achieve rapid matching and similarity judgment between different observation information, it is necessary to perform feature compression representation on the target region and map it into a fixed-length binary code.

[0038] Let the observation information of the target area be ,in Represents pixel coordinates, This represents the corresponding pixel value.

[0039] First, the target area is normalized in size and converted to grayscale to eliminate the influence of scale and color differences.

[0040] The processed observation information is then subjected to frequency domain transformation to extract its low-frequency feature components. As a specific implementation, Discrete Cosine Transform (DCT) can be used to transform the observation information from the spatial domain to the frequency domain: ; in, This represents the frequency domain coefficients, and M represents the total number of rows or columns in the pixel array. , Indicates frequency index, Representing cosine similarity, this transformation converts observation information from the spatial domain to the frequency domain, where the low-frequency component reflects the overall structural features of the target. Select low-frequency coefficients in the frequency domain to form a set L, and calculate its average value: ; in, K represents the average value of the coefficients, and K represents the number of low-frequency coefficients in set L. Each low-frequency coefficient is compared to the average value to generate a binary code: ; in, When the corresponding frequency component is stronger than the average level, it reflects that the structural feature is more significant in the current observation information; when This indicates that the component is weaker than the average level; All binary codes are arranged in a predetermined order to form a fixed-length binary sequence. , which serves as the perceptual hash value for the target region.

[0041] Furthermore, for the perceptual hash values ​​corresponding to different observation information, the Hamming distance can be calculated. Similarity measurement is performed; the smaller the distance, the more similar the observed information. , These represent the first observation information. Second observation information The frequency component corresponding to the e-th low-frequency coefficient.

[0042] This embodiment also provides a distributed data transmission system for unmanned systems based on fuzzy belief and hysteresis triggering, the system comprising: The sensor module is used to collect target observation information of the unmanned system at the current moment; A feature extraction coprocessor is used to extract the uncertainty feature vector and mission value feature vector of the unmanned system in parallel. The fuzzy inference computing unit is used to perform two-layer fuzzy inference on the uncertainty feature vector and the task value feature vector, and output the comprehensive confidence level of the unmanned system at the current moment. The hysteresis trigger decision is used to calculate the high trigger threshold and low silence threshold based on the comprehensive confidence of the unmanned system at past times; and to update the data transmission status of the unmanned system at the current time based on the hysteresis comparison result between the comprehensive confidence of the unmanned system at the current time and the high trigger threshold or low silence threshold. The differential coding engine is used to enable the unmanned system to execute a differential data transmission strategy based on the corresponding comprehensive confidence level when the updated data transmission status is active. The communication module is used to enable network communication between unmanned systems; All of the above modules can be integrated into the embedded system of the unmanned system.

[0043] It is worth noting that the unmanned systems described in this embodiment include, but are not limited to, drones, unmanned vehicles, unmanned boats, and other systems that can complete designated tasks with no or minimal human intervention.

[0044] The distributed data transmission method for unmanned systems based on fuzzy belief and hysteresis triggering provided in this embodiment has the following beneficial effects: Through the synergy of S1 to S3, a closed-loop technology system of value perception, anti-interference decision-making, and precise delivery is formed. First, the two-layer fuzzy inference in S1 is not a simple weighting, but rather dynamically reshapes the connotation of data confidence based on task value, ensuring that high-value information is not filtered by rigid standards. Second, the hysteresis triggering mechanism in S2 is not a simple double threshold; its adaptive characteristics ensure that the triggering decision can resist data fluctuations and respond to the global network state. Finally, the differentiated data transmission strategy in S3 directly inherits the output results of the first two steps, achieving a precise match between communication resources and information utility. The synergistic effect of these three elements jointly solves the problem of reliable, real-time transmission of high-value, low-quality data in low-bandwidth dynamic environments.

[0045] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0046] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A distributed data transmission method for unmanned systems based on fuzzy confidence and hysteresis triggering, characterized in that, include: S1: For any target observation information collected by an unmanned system at the current moment, extract the uncertainty feature vector and task value feature vector corresponding to the unmanned system in parallel; and output the comprehensive confidence level of the unmanned system at the current moment by passing the uncertainty feature vector and task value feature vector through two-layer fuzzy inference. S2: Calculate the high trigger threshold and low silence threshold based on the comprehensive confidence of the unmanned system at past times; update the data transmission status of the unmanned system at the current time based on the hysteresis comparison result between the comprehensive confidence of the unmanned system at the current time and the high trigger threshold or low silence threshold. S3: When the updated data transmission status is active, the unmanned system executes a differentiated data transmission strategy based on the corresponding comprehensive confidence level.

2. The distributed data transmission method for unmanned systems according to claim 1, characterized in that, The target observation information includes images, point clouds, or radio frequency signals.

3. The distributed data transmission method for unmanned systems according to claim 1, characterized in that, Extract the uncertainty feature vector of the unmanned system, including: The quality index is obtained by calculating the structural similarity index of the target observation information; The occlusion ratio of the target observation information is calculated using a segmentation algorithm; Estimating motion ambiguity of unmanned systems based on inertial measurement unit data; Calculate the signal-to-noise ratio of the target observation information.

4. The distributed data transmission method for unmanned systems according to claim 3, characterized in that, Extracting the task value feature vector of the unmanned system, including: The unmanned system acquires the task and calculates the normalized distance between the unmanned system and the target point in the task. Query the urgency of the retrieved tasks; Calculate the time-dependent decay factor based on the discovery duration.

5. The distributed data transmission method for unmanned systems according to claim 3, characterized in that, The uncertainty feature vector and the task value feature vector are subjected to two-layer fuzzy reasoning, including: Each element in the task value feature vector is fuzzified into its corresponding linguistic value through a membership function, and each linguistic value corresponding to the task value feature vector is mapped into a rule modulation factor according to a preset mapping table. Based on the rule modulation factor and the preset fuzzy mapping rule, the uncertainty feature vector is mapped to the comprehensive confidence level of the unmanned system at the current moment.

6. The distributed data transmission method for unmanned systems according to claim 5, characterized in that, The preset fuzzy mapping rules are: If the quality index is a first preset value, the occlusion ratio is a second preset value, the motion blur is a third preset value, and the signal-to-noise ratio is a fourth preset value, then the value calculated by the regular modulation factor will be used as the overall confidence level.

7. The distributed data transmission method for unmanned systems according to claim 1, characterized in that, Based on the comprehensive confidence level of the unmanned system at past times, high trigger thresholds and low silence thresholds are calculated, including: A confidence sliding window of length W is set, and the confidence sliding window slides and samples the comprehensive confidence of the unmanned system at consecutive past moments with a preset step size. Calculate the mean and standard deviation of each composite confidence level in the confidence level sliding window corresponding to the current time. The high trigger threshold and low silence threshold are calculated based on the mean and the standard deviation, and the calculation formula includes: ； ； in, Indicates a high trigger threshold. Indicates a low silence threshold. This represents the mean of the overall confidence level. The standard deviation represents the overall confidence level. Indicates the first positive coefficient. Indicates the second positive coefficient. Indicates the third positive coefficient. This represents the network load factor at the current moment.

8. The distributed data transmission method for unmanned systems according to claim 7, characterized in that, The network load factor at the current moment includes the channel occupancy rate obtained by the unmanned system based on channel sensing at the current moment.

9. The distributed data transmission method for unmanned systems according to claim 7, characterized in that, Update the data transmission status of the unmanned system at the current moment, including: Obtain the data transmission status of the unmanned system at the current moment. If the comprehensive confidence level of the unmanned system at the current moment is greater than the high trigger threshold, and the data transmission status of the unmanned system at the current moment is silent, trigger data transmission and update the data transmission status of the unmanned system at the current moment to active. If the overall confidence level of the unmanned system at the current moment is less than the low silence threshold, and the data transmission status of the unmanned system at the current moment is active, then stop data transmission and update the data transmission status of the unmanned system at the current moment to silent. In all other cases, the data transmission status of the unmanned system remains unchanged at the current moment.

10. The distributed data transmission method for unmanned systems according to claim 4, characterized in that, S3 include: When the updated data transmission status is active, if the overall confidence level of the unmanned system is greater than the first interval threshold, a lightweight video encoder is used to compress the target observation information collected by the unmanned system into the first data packet. If the overall confidence level of the unmanned system is greater than the second interval threshold and less than or equal to the first interval threshold, then the low-quality region in the target observation information is located according to the occlusion ratio and motion ambiguity. The super-resolution network is called to enhance the low-quality region in real time. The enhanced region is then stitched together with the other high-quality regions and compressed to obtain compressed data. A binary mask is generated to identify the enhanced region. The compressed data and the binary mask are then packaged into a second data packet. If the overall confidence level of the UAV system is less than or equal to the second interval threshold and the urgency of the task is greater than the urgency threshold, then the target category and bounding box in the target observation information are extracted through the lightweight target detection network, and the perceptual hash value of the target area is calculated. The timestamp, target category, bounding box, perceptual hash value and re-observation request flag are encapsulated into a third data packet. The first data packet, the second data packet, or the third data packet corresponding to the unmanned system is broadcast to the neighboring unmanned system through a distributed communication protocol.

Images