Collaborative sensing method and device for unmanned aerial vehicle assisted sparse crowd sensing

Abstract

A kind of unmanned aerial vehicle auxiliary sparse crowd-sensing-oriented collaborative sensing method and device, method includes: using unmanned aerial vehicle to carry out sensing data collection to the sub-region in the sensing area, the participant is evaluated for sensing quality, obtain first sensing quality;Using the participant who has been evaluated to carry out sensing data collection, the participant who has not been evaluated is evaluated for sensing quality, obtain second sensing quality and participant dependency;Using the sensing data collected by the participant who has been evaluated, the participant who has been evaluated is evaluated for sensing quality, according to participant dependency, update sensing quality, obtain participant sensing quality;Data weighting fusion is carried out to sensing data, obtain input data;Using bayesian compressive sensing algorithm to calculate input data, determine the sensing data corresponding to the sub-region not sensed.The present application improves the accuracy of prediction result, reduces the influence of real physical world participant low-quality data on reasoning algorithm.

Description

Technical Field

[0001] This invention relates to the field of sparse swarm intelligence sensing technology, and more particularly to a collaborative sensing method and device for UAV-assisted sparse swarm intelligence sensing. Background Technology

[0002] With the rapid development of information technology and the widespread adoption of sensor-equipped smartphones, Sparse Mobile Crowd Sensing (SparseMCS) has become a promising emerging paradigm. This technology collects data from a select few sub-regions and leverages the inherent correlations between sensor data and data inference algorithms to deduce data from other sub-regions, thereby achieving perception of large-scale urban environments at a relatively low cost.

[0003] Sparse mobile swarm sensing technology offers several advantages. First, by rationally selecting sub-regions and employing sophisticated data inference algorithms, it can significantly reduce sensing costs and energy consumption, thereby improving system efficiency and sustainability. Second, this technology has broad application prospects in areas such as urban sensing and environmental monitoring. It can achieve large-scale, real-time sensing of various aspects of the city, providing valuable data support for decision-makers and researchers, and promoting the intelligent and sustainable development of cities.

[0004] Therefore, the rational selection of sub-regions and the improvement of inference algorithm accuracy play a crucial role in perception quality. However, existing technical solutions are often overly idealistic, neglecting the challenges and limitations of real-world scenarios. Current solutions generally assume that a sufficient number of users participate in perception in each region, and that each user's perception data is highly accurate. However, in real physical environments, the accuracy of perception data often falls short of expectations. Summary of the Invention

[0005] To address the problems existing in the prior art, the main objective of this invention is to provide a collaborative perception method and device for UAV-assisted sparse swarm intelligence perception, which improves the accuracy of perception result prediction for unperceived sub-regions and reduces the impact of low-quality data from real-world participants on inference algorithms.

[0006] To achieve the above objectives, embodiments of the present invention provide a collaborative perception method for UAV-assisted sparse swarm intelligence perception, the method comprising:

[0007] Using drones, perception data is collected in sub-regions of the perception area, and the perception quality of participants in the sub-regions is assessed to obtain the first perception quality of the assessed participants.

[0008] While using drones to collect perception data, perception data is also collected using participants who have already been evaluated, and perception quality is assessed for participants who have not been evaluated to obtain a second perception quality and determine participant dependencies.

[0009] Using the perceptual data collected by the evaluated participants, the perceptual quality of the evaluated participants is assessed, and the first and second perceptual qualities are updated according to the participants' dependencies to obtain the participants' perceptual quality.

[0010] Based on the participants' perceived quality, the collected perceptual data is weighted and fused to obtain the input data;

[0011] The Bayesian compressed sensing algorithm is used to calculate the input data and determine the sensing data corresponding to the unsensitized sub-regions within the sensing area.

[0012] Optionally, in one embodiment of the present invention, a drone is used to collect perception data in a sub-region of the perception area, and the perception quality of participants in the sub-region is assessed to obtain the first perception quality of the assessed participants, including:

[0013] The first perception data is obtained by using sensors deployed on the drone to collect perception data of the sub-region.

[0014] The system receives participant perception data sent by participants to be evaluated in the sub-region, and determines the first perception quality and degree of perception of the evaluated participants based on the first perception data and the participant perception data.

[0015] Optionally, in one embodiment of the present invention, the perceived quality of the evaluated participants is assessed using the perceptual data collected by the evaluated participants, and the first and second perceptual qualities are updated according to the participants' dependencies, resulting in the participant's perceived quality including:

[0016] When the values ​​of the first or second perceived quality of two evaluated participants are not equal, the participant with the larger value of the first or second perceived quality is used to collect perceived data, and the perceived quality of the other participant is evaluated to obtain the third perceived quality, and the participant dependency relationship is updated.

[0017] Based on the third perceived quality, the perceived quality and perceived level of the participants being evaluated are updated.

[0018] Based on the updated participant dependencies and the perceived quality of the participants being evaluated, the perceived quality and perceived level of participants who depend on the participants being evaluated are updated.

[0019] Optionally, in one embodiment of the present invention, the method further includes:

[0020] Obtain the drone's location and energy, and obtain the participants' locations and energy within the sub-region;

[0021] Based on the participant dependencies, determine the participant number that will be evaluated last for each participant;

[0022] Based on the drone's location, drone energy, participant's location, participant energy, participant ID, and participant's perceived quality, a multi-agent reinforcement learning algorithm is used to determine the drone's actions and the participants' actions.

[0023] This invention also provides a collaborative sensing device for UAV-assisted sparse swarm intelligence sensing, the device comprising:

[0024] The drone module is used to collect perception data in sub-regions of the perception area using drones, and to evaluate the perception quality of participants in the sub-regions to obtain the first perception quality of the evaluated participants.

[0025] The perception data module is used to collect perception data using drones, collect perception data using already evaluated participants, and evaluate the perception quality of unevaluated participants to obtain a second perception quality and determine participant dependencies.

[0026] The perceived quality module is used to evaluate the perceived quality of the evaluated participants using the perceived data collected by the evaluated participants, and to update the first and second perceived quality according to the participants' dependencies to obtain the participants' perceived quality.

[0027] The data fusion module performs weighted fusion of the collected perceptual data based on the participants' perceived quality to obtain the input data;

[0028] The data prediction module is used to calculate the input data using the Bayesian compressed sensing algorithm to determine the sensing data corresponding to the unsensitized sub-regions within the sensing area.

[0029] Optionally, in one embodiment of the present invention, the drone module includes:

[0030] The data collection unit is used to collect perception data of a sub-region using sensors deployed on the drone to obtain the first perception data;

[0031] The perception quality unit is used to receive participant perception data sent by the participants to be evaluated in the sub-region, and determine the first perception quality and the degree of perception of the evaluated participants based on the first perception data and the participant perception data.

[0032] Optionally, in one embodiment of the present invention, the quality sensing module includes:

[0033] The dependency unit is used to collect perception data from the participant with the larger value of the first or second perception quality when the values ​​of the two evaluated participants are not equal, and to evaluate the perception quality of the other participant to obtain the third perception quality and update the participant dependency relationship.

[0034] The first update unit is used to update the perceived quality and the degree of perception of the participants being evaluated based on the third perceived quality.

[0035] The second update unit is used to update the perceived quality and perceived level of participants who depend on the participants being evaluated, based on the updated participant dependencies and the perceived quality of the participants being evaluated.

[0036] Optionally, in one embodiment of the present invention, the apparatus further includes:

[0037] The location acquisition module is used to acquire the drone's location and energy, as well as the location and energy of participants in the sub-region;

[0038] The participant numbering module is used to determine the final participant number for each participant based on participant dependencies.

[0039] The action generation module is used to determine the drone actions and participant actions based on the drone's position, drone energy, participant's position, participant energy, participant ID, and participant's perceived quality using a multi-agent reinforcement learning algorithm.

[0040] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-described method.

[0041] The present invention also provides a computer-readable storage medium storing a computer program that performs the above-described methods by a computer.

[0042] The present invention also provides a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps of the above-described method.

[0043] This invention utilizes drone-assisted mobile crowd perception data, combined with an assessment of participants' perception quality. While evaluating participant quality, perception data is collected, and a data fusion algorithm is used to fuse data of different perception qualities. The fused data serves as input to a data inference algorithm, thereby predicting data in unperceived areas. This improves the accuracy of the input data for the data inference algorithm, enhances the accuracy of the prediction results, and reduces the impact of low-quality data from real-world participants on the inference algorithm. Attached Figure Description

[0044] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments will be 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.

[0045] Figure 1 This is a flowchart of a collaborative perception method for UAV-assisted sparse swarm intelligence perception according to an embodiment of the present invention;

[0046] Figure 2 This is a flowchart illustrating the perceived quality in an embodiment of the present invention;

[0047] Figure 3 This is a flowchart of the perceived quality update in an embodiment of the present invention;

[0048] Figure 4 This is a flowchart of the action generation process in an embodiment of the present invention;

[0049] Figure 5 This is a flowchart of a collaborative sensing method in a specific embodiment of the present invention;

[0050] Figure 6 This is a schematic diagram of the structure of a collaborative sensing device for sparse swarm intelligence sensing assisted by unmanned aerial vehicles (UAVs) according to an embodiment of the present invention.

[0051] Figure 7 This is a schematic diagram of the structure of the unmanned aerial vehicle module in an embodiment of the present invention;

[0052] Figure 8 This is a schematic diagram of the sensing quality module structure in an embodiment of the present invention;

[0053] Figure 9 This is a schematic diagram of the structure of a collaborative sensing device in another embodiment of the present invention. Detailed Implementation

[0054] This invention provides a collaborative sensing method and apparatus for unmanned aerial vehicle-assisted sparse swarm intelligence sensing.

[0055] 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.

[0056] like Figure 1 The diagram shows a flowchart of a collaborative perception method for UAV-assisted sparse crowd sensing according to an embodiment of the invention. The execution entities of the collaborative perception method for UAV-assisted sparse crowd sensing provided in this embodiment include, but are not limited to, computers, cloud service platform servers, etc. This invention uses UAV-assisted mobile crowd sensing data, combined with an assessment of the participants' perception quality, to collect perception data while assessing participant quality. A data fusion algorithm is used to fuse data of different perception qualities, and the fused data is used as input to a data inference algorithm. This predicts data in unperceived areas, improves the accuracy of the input data for the data inference algorithm, improves the accuracy of the prediction results, and reduces the impact of low-quality data from real-world participants on the inference algorithm. The method shown in the diagram includes:

[0057] Step S1: Use a drone to collect perception data in a sub-region of the perception area and evaluate the perception quality of the participants in the sub-region to obtain the first perception quality of the evaluated participants.

[0058] Step S2: While collecting perception data using drones, collect perception data using the participants who have been evaluated, and evaluate the perception quality of the unevaluated participants to obtain the second perception quality and determine the participant dependencies.

[0059] Step S3: Using the perceptual data collected by the evaluated participants, assess the perceptual quality of the evaluated participants, and update the first and second perceptual qualities according to the participants' dependencies to obtain the participants' perceptual quality.

[0060] Step S4: Based on the participants' perceived quality, the collected perceptual data is weighted and fused to obtain the input data;

[0061] Step S5: Calculate the input data using the Bayesian compressed sensing algorithm to determine the sensing data corresponding to the unsensitized sub-regions within the sensing area.

[0062] Among these methods, calibrated high-precision sensors are deployed on drones, and the cloud service platform evaluates the perception quality of participants by observing the intersection of drones and sparse swarm intelligent sensing participants.

[0063] Furthermore, during this period, the data perceived by the drone in different areas is uploaded to the cloud service platform. Participants whose data has been calibrated by the drone can update the perception quality of other participants, and the perception data will also be uploaded to the cloud service platform.

[0064] Furthermore, a data fusion method is employed to obtain more accurate data for the perceived data of participants with different perceived qualities within the same area. The cloud service platform uses inference algorithms to infer data from unperceived sub-regions based on the already uploaded regional data.

[0065] Specifically, at the initial moment, the cloud service platform controls the drone's movements to collect perception data from sub-regions within the perception area and assess the perception quality of participants within those sub-regions. The perception area comprises multiple pre-defined sub-regions.

[0066] In this process, the cloud service platform utilizes perception data collected by drones to assess the perception quality of the participants. Specifically, the cloud service platform receives perception data sent by the participants and, by combining the drone-collected perception data with the participants' perception data, obtains their corresponding initial perception quality. This process can be referred to as assessing the perception quality of participants using perception data collected by drones. Specifically, the initial perception quality here refers to the perception quality obtained through the perception quality assessment using the perception data collected by drones.

[0067] Furthermore, when a participant that has been evaluated by the drone appears in a sub-region, the cloud service platform uses the evaluated participant to collect perception data in the sub-region of the perception area and performs perception quality assessment on the unevaluated participants in the sub-region. This process can be called using the perception data collected by the evaluated participants to perform perception quality assessment on the unevaluated participants, and the resulting perception quality is the second perception quality.

[0068] Specifically, when the perceived quality assessment of an unassessed participant is conducted by a participant who has already been assessed, it is determined that the assessed participant depends on the participant who assessed their perceived quality, thus obtaining participant dependency.

[0069] In this system, after the drone collects perception data and assesses the perception quality of the first participant, that participant can immediately begin collecting perception data to assess the perception quality of other unassessed participants. Simultaneously, the drone continues to collect perception data from sub-regions within the perception area and assess the perception quality of other unassessed participants. By simultaneously conducting perception quality assessments with the drone and already assessed participants, the number of participants assessing other unassessed participants increases as more participants are evaluated, thus significantly improving the efficiency of perception quality assessment.

[0070] As one embodiment of the present invention, such as Figure 3 As shown, using the perceptual data collected by the evaluated participants, the perceived quality of the participants is assessed. Based on participant dependencies, the first and second perceptual qualities are updated, resulting in the participants' perceived quality including:

[0071] Step S31: When the values ​​of the first or second perceived quality of two evaluated participants are not equal, use the participant with the larger value of the first or second perceived quality to collect perceived data, evaluate the perceived quality of the other participant, obtain the third perceived quality, and update the participant dependency relationship.

[0072] Step S32: Based on the third perceived quality, update the perceived quality and perceived level of the participants being evaluated.

[0073] Step S33: Based on the updated participant dependencies and the perceived quality of the participants being evaluated, update the perceived quality and perceived level of the participants who depend on the participants being evaluated.

[0074] In this process, when both participants are already evaluated, the cloud service platform compares their perceived quality, which can be either a first perceived quality or a second perceived quality. Using the participant with the higher perceived quality value, the platform collects perceived data from the participant with the lower perceived quality value and receives the perceived data sent by the participant with the lower perceived quality value. The collected perceived data and the perceived data sent by the participant with the lower perceived quality value are then used to evaluate the perceived quality, resulting in a third perceived quality. This process can be described as collecting perceived data from already evaluated participants and then evaluating their perceived quality. It should be noted that the terms "first," "second," and "third" in this invention do not have specific meanings and are only used for distinction.

[0075] Specifically, when two participants evaluate each other's perceived quality, it indicates that the participant being evaluated has already conducted a second evaluation. Therefore, the obtained third perceived quality replaces the previous perceived quality. Furthermore, if the evaluated participant has a dependency relationship, that dependency is updated to depend on the participant who most recently evaluated them.

[0076] Furthermore, when a participant's perceived quality is updated, the perceived quality of all participants who depend on that participant must also be updated.

[0077] It should be noted that the quality assessment in this invention does not stop at the third perceived quality, that is, the mutual assessment process between the assessed participants. The perceived quality of the assessed party can be the updated third perceived quality, thereby updating its perceived quality again. This continuous updating and iteration continuously improves the accuracy of the assessment of the participants' perceived quality.

[0078] In this embodiment, by setting a preset perception period, the process of collecting perception data and evaluating perception quality of participants is completed, thereby obtaining the perception quality of participants and the degree to which participants are perceived.

[0079] Furthermore, each time a participant is evaluated, their perceived level is updated; for example, if the initial perceived level is 0, it is updated to 1, and so on. Specifically, the perceived level is used to calculate the participant's perceived quality.

[0080] In this embodiment, the cloud service platform utilizes the perceived quality of each participant to perform weighted fusion of the collected perceived data, thereby obtaining the input data. Specifically, the weighted fusion is performed using the following formula:

[0081]

[0082] Where z is the number of participants uploading sensing data in the current area, and q j To provide participants with a perceived quality. For participants' perception data.

[0083] Furthermore, the input data is the input data for data inference algorithms, such as the Bayesian compressed sensing algorithm, which enables the prediction of sensing data for unsensed sub-regions within the sensing area.

[0084] As one embodiment of the present invention, such as Figure 2 As shown, a drone is used to collect perception data in a sub-region of the perception area, and the perception quality of participants in the sub-region is assessed. The first perception quality of the assessed participants includes:

[0085] Step S11: Use sensors deployed on the drone to collect perception data of the sub-region to obtain the first perception data;

[0086] Step S12: Receive participant perception data sent by the participants to be evaluated in the sub-region, and determine the first perception quality and degree of perception of the evaluated participants based on the first perception data and the participant perception data.

[0087] The cloud service platform receives perception data collected by sensors on the drone and uses it as the first perception data. It also receives perception data from participants in the sub-region to be evaluated. Using the first perception data and the participant perception data, the platform performs a perception quality assessment on the participants to be evaluated, obtaining their first perception quality. Thus, the participant to be evaluated completes the perception quality assessment and becomes an evaluated participant. Specifically, the perception quality assessment is performed using the following formula:

[0088]

[0089] Among them, R i It is the participants' perceived quality coefficient. x i This represents the participant w. i The level of perception of participants was updated after the participants were assessed by the drone:

[0090] x i =x i +1 (3)

[0091] Where, initially, x i =0, C i This represents the discrepancy between participant data and drone data. GT represents the drone's sensor data. This represents the average value of a single sensing data point from a drone or participant.

[0092]

[0093] Furthermore, Variance of participant data:

[0094]

[0095] As one embodiment of the present invention, such as Figure 4 As shown, the method also includes:

[0096] Step S41: Obtain the drone's location and energy, and obtain the participants' locations and energy in the sub-region;

[0097] Step S42: Based on the participant dependency relationship, determine the participant number that will be evaluated by each participant in the end;

[0098] Step S43: Based on the drone's location, drone energy, participant's location, participant energy, participant ID, and participant's perception quality, a multi-agent reinforcement learning algorithm is used to determine the drone's actions and the participants' actions.

[0099] In order to improve the adaptability of data collectors to self-organize and cooperate in dynamic environments, this invention adopts a model based on multi-agent reinforcement learning.

[0100] Furthermore, the multi-agent reinforcement learning algorithm can specifically adopt the MADDPG algorithm, in which the drone and the participant are respectively considered as agents.

[0101] Specifically, by combining the drone's location, drone energy, participant's location, participant energy, and participant ID with the obtained participant perceived quality, the MADDPG algorithm is used to calculate the drone's actions and the participant's actions.

[0102] The drone's actions include waiting in place, flying up, down, left, right, etc., while the participants' actions include both perceived and unperceived data.

[0103] The collaborative perception method for drone-assisted sparse crowd perception provided by this invention considers the perception quality of participants. It introduces perception data of mobile crowds assisted by drones equipped with high-precision sensors and uses a participant data quality assessment method to upload perception data while assessing the quality of participants. When the data is uploaded to the cloud service platform, the platform fuses data of different perception qualities according to a data fusion algorithm. The fused data is used as input to a data inference algorithm to obtain data of unperceived areas.

[0104] Furthermore, by introducing a multi-agent reinforcement learning algorithm for sub-region selection using drones, this invention can provide a better sub-region selection scheme in the real physical world. In addition, because this invention considers the evaluation of participant quality, the accuracy of the input data of the data inference algorithm is improved, thereby improving the accuracy of the result and reducing the impact of low-quality data from real-world participants on the inference algorithm.

[0105] This invention can effectively improve the quality of sensor data while effectively balancing personnel recruitment costs, thereby providing more optimized task allocation decisions. It can effectively utilize the sensor data of low-cost sensors of sparse swarm intelligence sensing participants, and is highly automated, highly adaptive, easy to deploy and implement, making it suitable for large-scale sparse swarm intelligence sensing activities.

[0106] In a specific embodiment of this invention, for the specific implementation of sparse swarm intelligence sensing, the reasonable selection of sub-regions and the improvement of the accuracy of the inference algorithm play a crucial role in the sensing quality. However, existing technical solutions are often too idealistic, neglecting the challenges and limitations in real-world scenarios. Therefore, in order to apply sparse swarm intelligence sensing to the real physical world, addressing the imbalance in the number and quality of participants in different regions of real-world scenarios is of paramount importance. The specific process of the collaborative sensing method for UAV-assisted sparse swarm intelligence sensing in this invention is as follows: Figure 5 As shown, it specifically includes:

[0107] S100 deploys calibrated high-precision sensors onto a drone, which collects perception data and assesses the participants' perception data when they pass by the participants.

[0108] S200: Participants who have been evaluated can evaluate other unevaluated participants and upload their perception data.

[0109] The S300 uploads the sensing data to the cloud service and uses a data fusion method to integrate the sensing data of participants with different sensing qualities in the same area. The fused data is used as input, and an inference algorithm is used to infer the data of the unsensitized area to obtain a more accurate sensing map.

[0110] S400 is a model based on multi-agent reinforcement learning that solves the task allocation process, enabling dynamic task allocation and assessment of participants' ability to perceive data, thereby providing high-quality sparse swarm intelligence perception results and optimized task allocation decisions.

[0111] Specifically, to implement a collaborative sensing method for UAV-assisted sparse swarm intelligence sensing, a platform must first be created to handle task publishing, data fusion, participant recruitment, and data inference. First, the sensing region needs to be defined and subdivided. For example, to know the real-time temperature of various locations in a certain region, the region is first designated as the sensing region and then divided into a set A of equal-sized sub-regions.

[0112] A = {a1, a2, ..., a} g}

[0113] Where g is the number of sub-regions, and the size of the sub-regions depends on the required data granularity. i Let g represent the i-th subregion in the set, satisfying 1≤i≤g.

[0114] Furthermore, the duration of the sensing activity is determined, and the entire sensing activity is divided into several sensing cycles, represented by a set T:

[0115] T = {t1, t2, ..., t} t}

[0116] Where t represents the number of sensing cycles in the entire sensing activity, t i Let represent the i-th sensing cycle in the set, satisfying 1 ≤ i ≤ t. In specific implementation, different sensing cycles can be set according to different needs. Then, a set of drones is introduced, represented by the following set U:

[0117] U = {u1, u2, ..., u} l}

[0118] Where l represents the total number of drones in the sparse swarm intelligence sensing system, and the number of drones can be adjusted automatically, u i Let i represent the i-th drone in the set, satisfying 1≤i≤l.

[0119] Furthermore, the participants are represented by the following set W:

[0120] W = {w1, w2, ..., w h}

[0121] Where h represents the total number of participants in the sparse swarm intelligence sensing system, and participants can choose users registered on the platform within the sensing area, w i Let represent the i-th participant in the set, satisfying 1≤i≤h.

[0122] The data samples from drones and participants' perceptions are as follows:

[0123] d i ={d i,1 ,...,d i,n}

[0124] Where n represents the total number of times the sensing drone and participants continuously sense each time, and the total number of sensing times can be set according to needs, d i,j Let represent the j-th data point in a certain perception of the i-th participant in the set, satisfying 1≤i≤h+l, 1≤j≤n.

[0125] When the perception activity begins, the perception quality of all participants is unknown; only the UAV collects perception data and evaluates the perception quality of the participants. When the UAV and the participants intersect in the same sub-region, the UAV and the participants in the current region collect perception data simultaneously. The perception quality of the participants can be evaluated using the perception data. The perception quality of the participants can be calculated using formulas (2), (4)-(5), and the degree to which the participants are perceived can be calculated using formula (3).

[0126] In this embodiment, after the quality of some participants has been evaluated, the participants can further evaluate each other, as follows:

[0127] S201: After drone assessment, there are participants who were assessed by drones, drones, and participants who were not assessed in the area;

[0128] S202: Participants evaluated by drones i You can evaluate participants who were not evaluated. j When participant w j was w i After evaluation, participant w was identified. j Depends on participant w i ;

[0129] S203: For the two evaluated participants w i and w j If R i ≠R j Then the side with the larger R value can be evaluated as having the smaller R value;

[0130] S204: If participant w i After being evaluated, it depends on w i All participants will be updated;

[0131] S205: If it exists in the dependent participant w i Participants w j Then repeat step S204;

[0132] Among them, participant w i Evaluation of participants w j Then the participant w j The degree of evaluation x j Update using the following formula:

[0133] x i =x i +Re i (6)

[0134] Among them, Re i Calculate using the following formula:

[0135] Re i = -log(1-R) i (7)

[0136] When participant w j The ability to perceive data is assessed when participant quality is again assessed by participants w i Assessment, and updates to perceived quality:

[0137]

[0138]

[0139]

[0140] in, and This represents the value from the last update.

[0141] In this embodiment, when a time slot ends, the platform receives sensor data from the participants and the drone, and fuses the sensor data using the following formula. The participant weight value can be calculated using formula (1) by considering the participant's perceived level and perceived quality.

[0142] The fused data is obtained by calculating the weight of each participant and then performing a weighted average based on those weights. This approach better utilizes the perceptual data of different participants and takes into account their differences in perceptual abilities, thereby improving the overall quality of data fusion.

[0143] In this embodiment, at the end of the current sensing cycle, the platform uses the acquired sub-region sensing data and a data inference algorithm to infer data from the unsensitized sub-regions. Specifically, Bayesian Compressive Sensing (BCS) can be used to reconstruct the sensor map. Bayesian Compressive Sensing (BCS) can be represented as:

[0144]

[0145]

[0146] Where F is the projection matrix, This represents the intrinsic noise variance of the noise r. Reconstructing the sensor map involves finding... satisfy:

[0147]

[0148] Therefore, by reconstructing the sensor map, the prediction results can be directly obtained, namely the sensing data of the unsensitized sub-regions.

[0149] In this embodiment, a multi-agent reinforcement learning algorithm is used as the solution algorithm for data collection and participant quality assessment. Specifically, the MADDPG algorithm can be adopted, where the drone and the participant are respectively considered agents, and their states are as follows:

[0150] S = {S1, S2, S3, S4}

[0151] Where S1 represents the positions of the participants and the drone, S1 is:

[0152]

[0153] Where S2 represents the participant's perceived quality. S2 is:

[0154] S2={q1,q2,...,q h}

[0155] Where S3 represents the drone's energy, i.e., its electrical charge, S3 is:

[0156]

[0157] Where S4 represents the participant ID of the last participant evaluated. When a participant is evaluated by a drone or not evaluated, their ID is themselves, and S4 is:

[0158] S4={de1,de2,...,de h}

[0159] The generated action is:

[0160]

[0161] Among them, a t This indicates the task assignment for the drone and the participants, where drone u i action These represent the actions of the participants: the drone chooses to stay in place, fly up, down, left, or right in each time slot. These represent the participants' perceived data and unperceived data, respectively.

[0162] Furthermore, the reward is:

[0163]

[0164] Where, r t This represents the weighted sum of the error in reconstructing the sensor map, the drone's energy, and the participant's reward.

[0165] In this embodiment, the multi-agent algorithm flow is as follows:

[0166] S401: The UAV collects perception data and assesses the quality of participant data. Based on the location of participants in the perception area, the currently perceived area on the perception map, the perception quality of participants in the perception area, and the participant dependencies, the UAV generates actions using the above-mentioned strategy function.

[0167] S402: Participants collect perception data and evaluate the quality of participant data. Based on the location of the drone, the currently perceived area on the perception map, the perception quality of participants within the perception area, and participant dependencies, participants use the above-mentioned policy function to generate actions.

[0168] S403: Drones and the participants being evaluated upload perception data to the platform, which then runs a data fusion algorithm to fuse data of different perception qualities within a sub-region;

[0169] S404: The platform uses inference algorithms to infer the perception data of unknown areas based on known perception data;

[0170] S405: Check the termination condition. If the predetermined number of rounds has not been reached, execute S401; otherwise, terminate the multi-agent reinforcement learning perception task optimization algorithm.

[0171] This invention fully considers the impact of uneven distribution of participants and imbalance in the perceived quality of participants in the real physical world, and adopts a multi-agent reinforcement learning algorithm to control task allocation to adapt to dynamic physical environments. This invention has a high degree of automation, strong adaptability, is easy to deploy and implement, and is more suitable for sensor data collection in real environments.

[0172] like Figure 6 The figure shows a schematic diagram of a collaborative sensing device for sparse swarm intelligence sensing assisted by unmanned aerial vehicles (UAVs) according to an embodiment of the present invention. The device shown in the figure includes:

[0173] The drone module 10 is used to collect perception data in a sub-region of the perception area using a drone, and to evaluate the perception quality of the participants in the sub-region to obtain the first perception quality of the evaluated participants.

[0174] The perception data module 20 is used to collect perception data using drones, collect perception data using the participants who have been evaluated, and evaluate the perception quality of the participants who have not been evaluated to obtain a second perception quality and determine the participants' dependencies.

[0175] The perceived quality module 30 is used to evaluate the perceived quality of the evaluated participants using the perceived data collected by the evaluated participants, and to update the first perceived quality and the second perceived quality according to the participant dependency relationship to obtain the participant's perceived quality.

[0176] Data fusion module 40 is used to perform data weighting and fusion on the collected perceptual data according to the participants' perceived quality to obtain input data;

[0177] The data prediction module 50 is used to calculate the input data using the Bayesian compressed sensing algorithm to determine the sensing data corresponding to the unsensitized sub-regions within the sensing area.

[0178] As one embodiment of the present invention, such as Figure 7 As shown, the drone module 10 includes:

[0179] Data collection unit 11 is used to collect perception data of a sub-area using sensors deployed on the drone to obtain first perception data;

[0180] The perception quality unit 12 is used to receive participant perception data sent by the participants to be evaluated in the sub-region, and to determine the first perception quality and the degree of perception of the evaluated participants based on the first perception data and the participant perception data.

[0181] As one embodiment of the present invention, such as Figure 8 As shown, the quality sensing module 30 includes:

[0182] The dependency relationship unit 31 is used to collect perception data using the participant with the larger value of the first or second perception quality when the values ​​of the first or second perception quality of two evaluated participants are not equal, and to evaluate the perception quality of the other participant to obtain the third perception quality and update the participant dependency relationship.

[0183] The first update unit 32 is used to update the perceived quality and the degree of perception of the participants being evaluated based on the third perceived quality.

[0184] The second update unit 33 is used to update the perceived quality and perceived level of participants who depend on the participants being evaluated, based on the updated participant dependency relationship and the perceived quality of the participants being evaluated.

[0185] As one embodiment of the present invention, such as Figure 9 As shown, the device also includes:

[0186] The location acquisition module 60 is used to acquire the drone's location and energy, and to acquire the location and energy of participants in the sub-region;

[0187] The participant numbering module 70 is used to determine the participant number that will be evaluated last for each participant based on the participant dependency relationship.

[0188] The action generation module 80 is used to determine the drone actions and the actions of the participants based on the drone's position, drone energy, participant's position, participant energy, participant number, and participant's perceived quality using a multi-agent reinforcement learning algorithm.

[0189] Based on the same concept as the aforementioned collaborative sensing method for UAV-assisted sparse swarm intelligence sensing, this invention also provides a collaborative sensing device for UAV-assisted sparse swarm intelligence sensing. Since the principle underlying this collaborative sensing device for UAV-assisted sparse swarm intelligence sensing is similar to that of the aforementioned collaborative sensing method, the implementation of this collaborative sensing device for UAV-assisted sparse swarm intelligence sensing can refer to the implementation of the aforementioned collaborative sensing method, and will not be repeated here.

[0190] This invention utilizes drone-assisted mobile crowd perception data, combined with an assessment of participants' perception quality. While evaluating participant quality, perception data is collected, and a data fusion algorithm is used to fuse data of different perception qualities. The fused data serves as input to a data inference algorithm, thereby predicting data in unperceived areas. This improves the accuracy of the input data for the data inference algorithm, enhances the accuracy of the prediction results, and reduces the impact of low-quality data from real-world participants on the inference algorithm.

[0191] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-described method.

[0192] The present invention also provides a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps of the above-described method.

[0193] The present invention also provides a computer-readable storage medium storing a computer program that performs the above-described methods by a computer.

[0194] The electronic device may also include: a communication module, an input unit, an audio processor, a display, and a power supply. It is worth noting that the electronic device does not necessarily need to include all of the above components; furthermore, the electronic device may include components not shown in the figures, which can be found in existing technologies.

[0195] A central processing unit, sometimes also called a controller or operating control, may include a microprocessor or other processor device and / or logic device, which receives input and controls the operation of various components of an electronic device.

[0196] The memory may be, for example, one or more of a cache, flash memory, hard drive, removable media, volatile memory, non-volatile memory, or other suitable devices. It may store the aforementioned failure-related information, and also store programs for executing that information. The central processing unit can execute the program stored in the memory to perform information storage or processing, etc.

[0197] An input unit provides input to the central processing unit. This input unit may be, for example, a button or touch input device. A power supply provides power to the electronic device. A display is used to display images and text. This display may be, for example, an LCD display, but is not limited to this.

[0198] The memory can be solid-state memory, such as read-only memory (ROM), random access memory (RAM), SIM card, etc. It can also be memory that retains information even when power is off, can be selectively erased, and contains more data; examples of this type of memory are sometimes called EPROM, etc. The memory can also be some other type of device. Memory includes buffer memory (sometimes called a buffer). Memory can include application / function storage for storing application programs and function programs or processes for performing operations of the electronic device via the central processing unit.

[0199] The memory may also include a data storage section for storing data, such as contacts, digital data, pictures, sounds, and / or any other data used by the electronic device. The memory's driver storage section may include various drivers for the electronic device's communication functions and / or for performing other functions of the electronic device (such as messaging applications, address book applications, etc.).

[0200] The communication module is a transmitter / receiver that sends and receives signals via an antenna. The communication module (transmitter / receiver) is coupled to the central processing unit to provide input signals and receive output signals, which can be the same as in a conventional mobile communication terminal.

[0201] Based on different communication technologies, multiple communication modules can be incorporated into the same electronic device, such as cellular network modules, Bluetooth modules, and / or wireless LAN modules. The communication module (transmitter / receiver) is also coupled to a speaker and microphone via an audio processor to provide audio output through the speaker and receive audio input from the microphone, thereby enabling typical telecommunications functions. The audio processor can include any suitable buffer, decoder, amplifier, etc. Furthermore, the audio processor is coupled to a central processing unit, enabling on-device recording via the microphone and on-device playback of stored sound via the speaker.

[0202] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0203] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0204] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0205] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0206] Specific embodiments have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A collaborative sensing method for UAV-assisted sparse swarm intelligence sensing, characterized in that, The method includes: Using drones, perception data is collected in sub-regions of the perception area, and the perception quality of participants in the sub-regions is assessed to obtain the first perception quality of the assessed participants. While collecting perception data using the drone, perception data is also collected using the evaluated participants, and perception quality is evaluated for the unevaluated participants to obtain a second perception quality and determine the participant dependencies. Using the perceptual data collected by the evaluated participants, the perceptual quality of the evaluated participants is evaluated, and the first and second perceptual qualities are updated according to the participant dependencies to obtain the participant's perceptual quality. Based on the participants' perceived quality, the collected perceptual data is weighted and fused to obtain the input data; The input data is calculated using a Bayesian compressed sensing algorithm to determine the sensing data corresponding to the unsensitized sub-regions within the sensing area.

2. The method according to claim 1, characterized in that, The method of using drones to collect perception data in sub-regions of the perception area and assessing the perception quality of participants in the sub-regions to obtain the first perception quality of the assessed participants includes: The first perception data is obtained by using sensors deployed on the drone to collect perception data of the sub-region. The system receives participant perception data sent by participants to be evaluated in the sub-region, and determines the first perception quality and degree of perception of the evaluated participants based on the first perception data and the participant perception data.

3. The method according to claim 2, characterized in that, The process of using the perceptual data collected from the evaluated participants to assess their perceptual quality, and updating the first and second perceptual qualities based on the participant dependencies, to obtain the participant's perceptual quality includes: When the values ​​of the first or second perceived quality of two evaluated participants are not equal, the participant with the larger value of the first or second perceived quality is used to collect perceived data, and the perceived quality of the other participant is evaluated to obtain the third perceived quality, and the participant dependency relationship is updated. Based on the third perceived quality, the perceived quality and perceived level of the participants being evaluated are updated. Based on the updated participant dependencies and the perceived quality of the participants being evaluated, the perceived quality and perceived level of participants who depend on the participants being evaluated are updated.

4. The method according to claim 1, characterized in that, The method further includes: Obtain the drone's location and energy, and obtain the participants' locations and energy in the sub-region; Based on the participant dependencies, determine the participant number that will be evaluated by each participant in the end; Based on the drone's location, drone energy, participant's location, participant energy, participant ID, and participant's perceived quality, a multi-agent reinforcement learning algorithm is used to determine the drone's actions and the participants' actions.

5. A collaborative sensing device for UAV-assisted sparse swarm intelligence sensing, characterized in that, The device includes: The drone module is used to collect perception data in sub-regions of the perception area using drones, and to evaluate the perception quality of participants in the sub-regions to obtain the first perception quality of the evaluated participants. The perception data module is used to collect perception data using the UAV, collect perception data using the evaluated participants, evaluate the perception quality of the unevaluated participants, obtain a second perception quality, and determine the participant dependencies. The perceived quality module is used to evaluate the perceived quality of the evaluated participants using the perceived data collected by the evaluated participants, and to update the first and second perceived quality according to the participant dependencies to obtain the participant's perceived quality. The data fusion module performs data weighting and fusion on the collected perception data based on the perceived quality of the participants to obtain the input data; The data prediction module is used to calculate the input data using a Bayesian compressed sensing algorithm to determine the sensing data corresponding to the unsensitized sub-regions within the sensing area.

6. The apparatus according to claim 5, characterized in that, The drone module includes: A data collection unit is used to collect perception data of a sub-region using sensors deployed on the UAV to obtain first perception data. The perception quality unit is used to receive participant perception data sent by the participants to be evaluated in the sub-region, and determine the first perception quality and the degree of perception of the evaluated participants based on the first perception data and the participant perception data.

7. The apparatus according to claim 5, characterized in that, The quality sensing module includes: The dependency unit is used to collect perception data from the participant with the larger value of the first or second perception quality when the values ​​of the two evaluated participants are not equal, and to evaluate the perception quality of the other participant to obtain the third perception quality and update the participant dependency relationship. The first update unit is used to update the perceived quality and the degree of perception of the participants being evaluated based on the third perceived quality. The second update unit is used to update the perceived quality and perceived level of participants who depend on the participants being evaluated, based on the updated participant dependencies and the perceived quality of the participants being evaluated.

8. The apparatus according to claim 5, characterized in that, The device further includes: The location acquisition module is used to acquire the drone's location and energy, and to acquire the location and energy of participants in the sub-region; The participant numbering module is used to determine the final participant number for each participant based on participant dependencies. The action generation module is used to determine the drone actions and participant actions based on the drone's position, drone energy, participant's position, participant energy, participant number, and participant's perceived quality using a multi-agent reinforcement learning algorithm.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method according to any one of claims 1 to 4.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that enables a computer to execute the method according to any one of claims 1 to 4.

11. A computer program product, comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method according to any one of claims 1 to 4.

Images