Drone path re-planning method and system fusing profile information

Abstract

This invention discloses a method and system for replanning unmanned aerial vehicle (UAV) paths by integrating profile information, relating to the field of UAV path planning technology. The method includes: acquiring initial profile information containing electromagnetic geometric data, electromagnetic signal prediction field, and uncertainty parameters; extracting local flight paths and mapping them to the electromagnetic signal prediction field; determining the communication threat level based on the acquired electromagnetic signal prediction distribution and uncertainty parameters; if the communication threat level is greater than a preset threshold, extracting physical patches as candidate reflection nodes; constructing a path topology map integrating physical space movement costs and electromagnetic link loss costs; performing path optimization search based on the path topology map; and generating a replanned path that includes reflection communication paths and / or active detection paths. Its beneficial effect is that it can overcome the dual dilemma of communication disruption and physical collision faced by UAVs in urban canyon environments.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) path planning technology, and in particular to a method and system for replanning UAV paths by integrating profile information. Background Technology

[0002] Urban areas with dense skyscrapers (i.e., "urban canyons") suffer from severe GNSS signal blockage, line-of-sight communication link disruption, and multipath reflection interference, making them high-risk airspace for UAV operations.

[0003] In handling flight missions in such complex environments, existing technologies generally treat "space physical obstacle avoidance" and "electromagnetic communication maintenance" as two isolated subsystems processed sequentially. Specifically, conventional three-dimensional path planning methods (such as A*, RRT*, or artificial potential field-based methods) only consider buildings as physical obstacles to be avoided, and their planning objectives are limited to collision avoidance and path shortestization within geometric space. Meanwhile, communication protection mechanisms typically only passively trigger failure protection actions such as hovering in place or returning to the original track in reverse order after the telemetry and control link signal is lost. This fragmented sequential control logic means that the "geometrically safe path" derived from path planning may happen to cross a communication blind spot, while the hovering or return actions triggered by communication protection may lock the UAV in an area where the signal cannot be restored, ultimately causing the UAV to fall into a double predicament of communication failure and physical collision in urban canyons. Summary of the Invention

[0004] In view of the above-mentioned prior art, this application is made. The embodiments of this application provide a method and system for replanning the path of an unmanned aerial vehicle (UAV) that integrates profile information. It can overcome the shortcomings of the prior art that treats "spatial physical obstacle avoidance" and "electromagnetic communication maintenance" as isolated subsystems processed sequentially, and overcome the dual dilemma of communication interruption and physical collision faced by UAVs in urban canyon environments.

[0005] According to one aspect of this application, a method for replanning a UAV path by fusing profile information is provided, comprising: acquiring initial profile information of the UAV's current pose, target waypoint, initial planned path, and flight airspace, wherein the initial profile information includes: electromagnetic geometry data containing multiple three-dimensional physical patches and their corresponding electromagnetic reflection properties, an electromagnetic signal prediction field pre-constructed based on the electromagnetic geometry data, and uncertainty parameters characterizing the reliability of the electromagnetic reflection properties; extracting a local flight path located ahead of the current pose in the initial planned path; mapping the local flight path to the electromagnetic signal prediction field to obtain an electromagnetic signal prediction distribution along the local flight path, and determining the local flight path based on the electromagnetic signal prediction distribution and the corresponding uncertainty parameters. The communication threat level of the flight path is determined; if the communication threat level is greater than a preset threshold, physical patches that meet preset reflection conditions are extracted from the electromagnetic geometry data as candidate reflection nodes, and a path topology graph is constructed based on the current pose of the UAV, the target waypoint, and each candidate reflection node. The connectivity edge cost of the path topology graph integrates the physical space movement cost and the electromagnetic link loss cost. Path optimization search is performed based on the path topology graph to generate and execute a replanning path. The replanning path includes: a reflection communication path that uses the target reflection node selected from each candidate reflection node as a signal reflection relay using the path optimization search results, and / or an active detection path aimed at reducing the uncertainty parameters in the flight airspace.

[0006] According to another aspect of this application, a UAV path replanning system integrating profile information is provided, comprising: an information acquisition module for acquiring the current pose, target waypoint, initial planned path, and initial profile information of the UAV, the initial profile information including: electromagnetic geometric data containing multiple three-dimensional physical patches and their corresponding electromagnetic reflection attributes, an electromagnetic signal prediction field pre-constructed based on the electromagnetic geometric data, and an uncertainty parameter characterizing the reliability of the electromagnetic reflection attributes; a path extraction module for extracting local flight paths located ahead of the current pose in the initial planned path; and a communication threat assessment module for mapping the local flight paths to the electromagnetic signal prediction field to obtain the electromagnetic signal prediction distribution along the local flight paths, and determining the communication threat assessment based on the electromagnetic signal prediction distribution and the corresponding uncertainty parameter. The system determines the communication threat level of the local flight path; a path topology construction module is used to determine whether the communication threat level is greater than a preset threshold. If so, it extracts physical patches that meet preset reflection conditions from the electromagnetic geometry data as candidate reflection nodes, and constructs a path topology graph based on the current pose of the UAV, the target waypoint, and each candidate reflection node. The connectivity edge cost of the path topology graph integrates the physical space movement cost and the electromagnetic link loss cost; a path replanning module is used to perform path optimization search based on the path topology graph to generate and execute a replanned path. The replanned path includes: a reflection communication path that uses the target reflection node selected from each candidate reflection node as a signal reflection relay using the path optimization search results, and / or an active detection path aimed at reducing the uncertainty parameters in the flight airspace.

[0007] According to another aspect of this application, an electronic device is provided, including a memory and a processor, the memory being used to store computer-executable instructions, and the processor being used to execute the computer-executable instructions, which, when executed by the processor, implement the steps of the method described above.

[0008] According to another aspect of this application, a computer-readable storage medium is provided that stores computer-executable instructions thereon, which, when executed by a processor, implement the steps of the method described above.

[0009] Compared with existing technologies, the UAV path replanning method and system based on the embodiments of this application, which integrates profile information, can overcome the shortcomings of existing technologies that treat "spatial physical obstacle avoidance" and "electromagnetic communication maintenance" as isolated subsystems processed sequentially. By binding the three-dimensional geometric features and electromagnetic reflection properties of the building facade into a unified physical patch to form initial profile information, and deeply integrating physical space movement costs and electromagnetic link loss costs in multi-domain joint path planning, the UAV can not only proactively predict communication threats during the path planning stage and transform physical obstacles into usable signal reflection relay resources to maintain communication links, but also achieve dynamic adaptation and topology self-healing in complex electromagnetic environments by combining active detection based on uncertain parameters and closed-loop correction mechanisms based on measured radio frequency data. This solves the dual dilemma of communication interruption and physical collision faced by UAVs in urban canyon environments. Attached Figure Description

[0010] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0011] Figure 1 This is a flowchart of the UAV path replanning method that integrates brief information according to the present invention.

[0012] Figure 2 This is a block diagram of the UAV path replanning system that integrates brief information according to the present invention.

[0013] Figure 3 This is a block diagram of an electronic device according to the present invention. Detailed Implementation

[0014] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0015] Exemplary method:

[0016] Figure 1 The illustration shows a method for replanning a drone path by fusing profile information according to an embodiment of this application, including steps S1 to S5.

[0017] like Figure 1As shown, in step S1, the current pose of the UAV, the target waypoint, the initial planned path, and the initial profile information of the flight airspace are obtained. The initial profile information includes: electromagnetic geometric data containing multiple three-dimensional physical patches and their corresponding electromagnetic reflection properties, electromagnetic signal prediction field pre-constructed based on electromagnetic geometric data, and uncertainty parameters characterizing the reliability of electromagnetic reflection properties.

[0018] In traditional 3D path planning, buildings typically participate in planning only as geometric envelopes (such as axis-aligned bounding boxes or simplified polyhedra), with the planning process focusing solely on the physical collision dimension of "whether a collision with a building is possible." However, in urban canyon environments, buildings are not only physical obstacles, but their facades are also critical electromagnetic boundaries affecting radio frequency (RF) signal propagation—the reflection, absorption, and scattering characteristics of RF signals vary greatly depending on the building surface material. If path planning relies solely on geometric information, it is impossible to predict whether a "geometrically safe" path will traverse communication blind spots, nor can it identify which building surfaces can be utilized as potential signal reflection relay resources. Therefore, this solution binds the 3D geometric features and electromagnetic reflection characteristics of buildings at the data structure level, forming unified initial profile information as the data foundation for all subsequent planning decisions.

[0019] Specifically, the electromagnetic geometry data consists of multiple three-dimensional physical patches, each corresponding to a geometric element on the building facade. Each physical patch is simultaneously bound to the geometric parameters of that element (including vertex coordinates, normal vector, and area) and electromagnetic reflection properties (including the reflection coefficient, incident angle response characteristics, and surface roughness level of the building material to which the element belongs in the working frequency band).

[0020] It's important to note that binding geometric parameters and electromagnetic reflection properties to the basic unit of a physical patch, rather than storing them separately in independent databases and then linking them through an index, is based on the following technical considerations: During reflection path planning, both geometric visibility determination (ray projection) and electromagnetic reflection feasibility assessment (reflection link loss calculation) need to be performed simultaneously on each candidate building surface. If geometric and electromagnetic data are stored separately, each assessment requires cross-database queries and data alignment, introducing unacceptable latency in airborne computing environments requiring real-time responses. The bound data structure of physical patches allows all the information needed for geometric determination and electromagnetic assessment to be obtained simultaneously in a single query.

[0021] The electromagnetic reflection properties of different building materials vary significantly. For example, Low-E coated glass curtain walls typically have a reflection coefficient of approximately 0.75 to 0.90 in the 2.4 GHz and 5.8 GHz frequency bands (due to the metal oxide layer in the coating forming a highly efficient reflector), classifying them as high-quality reflective surfaces. Concrete walls typically have a reflection coefficient of approximately 0.30 to 0.50 (primarily absorption, with surface roughness leading to diffuse reflection), classifying them as low-quality reflective surfaces. Aluminum alloy composite panel exterior walls and ordinary glass curtain walls fall somewhere in between. These differentiated electromagnetic reflection properties are fully recorded through the binding of physical panels to the corresponding building materials.

[0022] The third key component of the initial profile information is the uncertainty parameter. The electromagnetic reflection properties of different physical patches originate from different data acquisition methods, resulting in significant differences in their reliability: physical patches calibrated based on on-site radio frequency measurements have high confidence, those inferred from material records in building archives have medium confidence, and those inferred from AI annotations or conservative default values ​​in street view images have only low confidence. The uncertainty parameter is precisely the numerical indicator used to quantify these differences in reliability.

[0023] Uncertainty parameters are not merely supplementary information, but core variables driving a series of decisions in path replanning—their specific role will be gradually revealed as each step unfolds.

[0024] The second component of the initial profile is the electromagnetic signal prediction field, which is pre-constructed based on electromagnetic geometry data. Specifically, using the geometric parameters and electromagnetic reflection properties of all physical patches in the electromagnetic geometry data as input, and the locations and antenna parameters of the ground control station and cellular base station as the source model, the signal propagation model of the flight airspace is performed using the ray tracing method. The ray tracing method simulates the reflection and diffraction paths of the radio frequency signal after it originates from the source on the surface of buildings, calculating the predicted signal strength at each spatial location in the flight airspace. During the modeling process, the reflection order is limited to first-order and second-order reflections; third-order and higher reflections contribute very little to the signal strength while the computational cost increases exponentially. The flight airspace is divided into a three-dimensional voxel grid with a preset side length. Each voxel stores the predicted signal strength value at that location, the main signal path type (line-of-sight direct, first-order reflection, second-order reflection, or diffraction), and parameters such as multipath delay spread.

[0025] It should be noted that the pre-construction of the electromagnetic signal prediction field is an offline process, completed before the flight mission begins, and the calculation results are pre-loaded into the UAV's onboard storage. This layered design of offline pre-computation and online lightweight query is a key architectural choice to resolve the contradiction between "limited onboard computing power" and "real-time electromagnetic field perception"—complete ray tracing calculations require a large amount of computing power and are not suitable for real-time execution during flight; however, by storing the pre-computation results in the form of voxel grids, the onboard computing unit only needs to perform a hash table lookup operation during flight to obtain the signal prediction value at any location, and the time for a single query can be controlled in the millisecond range.

[0026] Furthermore, the prediction uncertainty of each voxel in the electromagnetic signal prediction field is calculated by propagating the uncertainty parameters of the physical patches that contribute the most to the prediction of that voxel's signal. Specifically, during the ray tracing pre-construction process, the predicted signal intensity of each voxel is composed of the superposition of the signal power of each signal path reaching that voxel (including the line-of-sight direct path and the reflection path via each physical patch), with each reflection path corresponding to a contributing physical patch. The power proportion of each contributing physical patch to the voxel's signal prediction can be calculated and recorded simultaneously during the pre-construction process. Based on this, the prediction uncertainty of the voxel is obtained by weighted summing of its uncertainty parameters according to the signal power proportion of each contributing physical patch—the larger the power proportion of a physical patch, the higher the weight of its uncertainty parameter in influencing the prediction uncertainty of the voxel. In a simplified implementation, the uncertainty parameter of the single physical patch that contributes the most to the voxel's signal prediction power can also be directly taken as the prediction uncertainty of that voxel. If the predicted signal intensity of a voxel mainly depends on the reflection contribution of a physical patch with a high uncertainty parameter, then the prediction uncertainty of that voxel itself is also correspondingly high—meaning that the predicted signal value of that voxel "may be inaccurate."

[0027] like Figure 1 As shown, in step S2, the local flight path located in front of the current pose in the initial planned path is extracted.

[0028] This step extracts a sequence of waypoints from the initial planned path, extending forward along the flight direction from the UAV's current pose for a preset time or distance, as a local waiting flight path. This local waiting flight path represents the route the UAV will soon fly over and is the target for communication threat assessment.

[0029] It should be noted that evaluating only a localized flight path rather than the entire initially planned path is based on two considerations: First, the prediction accuracy of the electromagnetic signal prediction field is related to its distance from the current pose—the farther the distance, the more accumulated environmental changes occur, and the lower the prediction's reference value. Second, prematurely replanning the path for distant segments may require repeated adjustments due to real-time corrections in the electromagnetic environment, resulting in unnecessary computational resource consumption. Therefore, this method adopts a rolling forward look-ahead strategy, focusing only on the localized flight path in the near future during each planning cycle. As the UAV continues to fly, the localized flight path is continuously updated and rolled forward.

[0030] like Figure 1 As shown, in step S3, the local flight path is mapped to the electromagnetic signal prediction field to obtain the electromagnetic signal prediction distribution along the local flight path, and the communication threat level of the local flight path is determined based on the electromagnetic signal prediction distribution and the corresponding uncertainty parameters.

[0031] Specifically, for each waypoint on the local flight path, the voxel at the corresponding position in the electromagnetic signal prediction field is queried, and the predicted signal strength value and prediction uncertainty of the voxel are read, thereby obtaining the electromagnetic signal prediction distribution along the entire local flight path.

[0032] The determination of the communication threat level is based on two dimensions: the first dimension is the absolute value of the predicted signal strength at each waypoint along the local flight path, and the second dimension is the uncertainty parameter corresponding to each waypoint.

[0033] When the predicted signal strength is higher than a preset safety threshold, but the uncertainty parameter exceeds a preset uncertainty threshold, the corresponding waypoint is identified as an area of ​​interest requiring active information acquisition. When the predicted signal strength is lower than a preset warning threshold, the corresponding waypoint is identified as a warning area requiring path replanning. In one implementation, the communication threat level is divided into three discrete levels according to severity, from low to high: normal level, attention level, and warning level. When the predicted signal strength is higher than the preset safety threshold and the uncertainty parameter is lower than the preset uncertainty threshold, it is identified as a normal level; when the predicted signal strength is higher than the preset safety threshold but the uncertainty parameter exceeds the preset uncertainty threshold, it is identified as a attention level; when the predicted signal strength is lower than the preset warning threshold, it is identified as a warning level regardless of the value of the uncertainty parameter. The communication threat level of a local waiting flight route is the highest threat level among all its waypoints. Accordingly, the preset threshold can be set as the attention level, meaning that when the communication threat level of a local waiting flight route reaches the attention level or the warning level, the path replanning process is triggered.

[0034] It's important to note that treating uncertainty parameters as an independent dimension for threat assessment, rather than relying solely on the absolute value of predicted signal strength, is a key feature that distinguishes this approach from existing technologies in communication threat assessment. In traditional approaches, if the predicted signal strength at a location is at a normal level, that location is considered secure for communication. However, when the electromagnetic reflection properties of the physical surface upon which the prediction relies are unreliable (e.g., the material data of a physical surface is derived solely from image annotations without actual radio frequency testing), the "normal" predicted signal strength at that location may deviate significantly from the actual situation—the predicted value indicates security, but actual communication interruption may occur when flying to that location. Uncertainty parameters are precisely the means to capture this kind of "seemingly secure but actually unreliable" risk.

[0035] like Figure 1 As shown, in step S4, it is determined whether the communication threat level is greater than a preset threshold. If so, physical patches that meet the preset reflection conditions are extracted from the electromagnetic geometry data as candidate reflection nodes. A path topology graph is constructed based on the current pose of the UAV, the target waypoint, and each candidate reflection node. The connectivity edge cost of the path topology graph integrates the physical space movement cost and the electromagnetic link loss cost.

[0036] When the determination of the communication threat level exceeds the preset threshold, the aircraft will no longer continue to fly along the initial planned path (which will cross the communication blind zone), but will instead initiate a reflection path replanning process.

[0037] The core idea of ​​reflection path replanning is: since the line-of-sight direct path between the ground control station and the UAV target flight area is blocked by buildings, are there any building facades that can reflect the signal emitted by the control station to a location reachable by the UAV, thus maintaining the communication link even when the line-of-sight path is unavailable? If such a reflective surface exists, the UAV only needs to adjust its flight path to ensure that it remains within the reflection coverage of a certain high-quality reflective surface during flight, thereby physically bypassing the path while maintaining the electromagnetic communication connection.

[0038] Based on this idea, candidate reflection nodes that satisfy preset reflection conditions are first selected from all physical patches in the electromagnetic geometry data. These preset reflection conditions include, but are not limited to, the following combined constraints:

[0039] First, geometric visibility constraints. Starting from the UAV's current pose, the ray casting method is used to determine which physical surfaces in the 3D building model are geometrically visible (i.e., there are no other buildings obstructing the view between the UAV and the physical surface). Simultaneously, it is necessary to determine whether the signal incident path between the signal source (ground control station or cellular base station) and the physical surface is unobstructed. Only physical surfaces that satisfy geometric visibility on both the UAV side and the signal source side meet the basic conditions for serving as reflection relays.

[0040] Second, electromagnetic reflection quality constraints. Based on the electromagnetic reflection properties of the physical surface, the total path loss of the reflection link for that physical surface is calculated, including the free-space propagation loss of the signal from the transmitter to the reflector, the reflection loss of the reflector itself (determined by the reflection coefficient and the angle of incidence), and the free-space propagation loss from the reflector to the UAV. Only physical surfaces with a predicted signal strength higher than a preset minimum communication threshold after reflection, plus a preset safety margin, are retained. In addition, the reflection coefficient of the physical surface must be higher than a preset lower limit for the reflection coefficient (to exclude low-quality reflective surfaces such as concrete and stone), and the signal incident angle must be within a preset effective incident angle range (reflection efficiency drops sharply at angles that are too small or too large).

[0041] Third, uncertainty constraints. For physical surfaces with excessively high uncertainty parameters in their electromagnetic reflection properties, even if their nominal reflection coefficient meets the aforementioned conditions, they are marked as requiring active detection and verification before they can be used as reliable reflection nodes, rather than being directly included in path planning. This constraint avoids basing path planning on unreliable reflection predictions.

[0042] After the above screening, physical patches that meet the preset reflection conditions are extracted as candidate reflection nodes.

[0043] Subsequently, a path topology graph is constructed based on the UAV's current pose, target waypoint, and candidate reflection nodes. In the path topology graph, the UAV's current pose and target waypoint serve as the start and end nodes, respectively, and each candidate reflection node serves as an intermediate node. If there is a line-of-sight path between two nodes without building obstruction, a connecting edge is established between them. The cost of the connecting edge in the path topology graph integrates the physical space movement cost and the electromagnetic link loss cost, specifically calculated using a multi-domain joint cost function, which includes:

[0044] A physical cost term used to characterize the energy consumption of a drone's geometrical flight or the distance it takes to avoid obstacles;

[0045] Electromagnetic loss cost term used to characterize the degree of signal attenuation;

[0046] An exploration penalty term that is positively correlated with the uncertainty parameter distributed along the connected edges; and

[0047] The exploration reward item is negatively correlated with the expected reduction of uncertainty parameters after the drone flies over the connected edge.

[0048] It should be noted that the exploration penalty and exploration reward terms in the aforementioned multi-domain joint cost function are the key mechanisms driving the path selection to achieve a dynamic balance between "conservative avoidance" and "active exploration." The exploration penalty term causes the path to tend to avoid areas with high uncertainty parameters—because high uncertainty means that signal prediction in that area may be inaccurate, and unexpected communication degradation may occur after flying in. The exploration reward term, on the other hand, gives a negative cost (i.e., a reward) to connected edges that "can significantly reduce the uncertainty parameters of the surrounding area after flying over"—because when the UAV flies over a certain area, the measured data collected by its onboard radio frequency front end can be used to calibrate the electromagnetic reflection properties of relevant physical patches in that area, thereby reducing uncertainty.

[0049] The expected reduction of uncertainty parameters in the exploration reward is calculated as follows: For a connected edge, determine the set of physical patches that the UAV can effectively observe while flying along the edge (i.e., physical patches that have a line-of-sight relationship with the connected edge and whose observation angle meets the preset minimum effective angle). Subtract the sum of the current uncertainty parameters of each physical patch in the set from the sum of the expected uncertainty parameters of each physical patch after measured calibration. The difference is the expected reduction of the connected edge. The expected uncertainty parameters after measured calibration are taken from a preset lower limit of confidence after measured calibration—this lower limit reflects the limit of uncertainty reduction achievable in a single flyby observation, determined by the measurement accuracy of the RF front-end and the flyby speed. The greater the difference between the current value of the uncertainty parameter and this lower limit, the greater the expected reduction and the higher the corresponding exploration reward.

[0050] The coexistence of these two cost terms produces a nontrivial path selection behavior: the path selection neither blindly ventures into high-uncertainty regions (because the exploration penalty imposes a cost) nor overly conservatively avoids all uncertain regions (because the exploration reward incentivizes paths to pass through locations on the edge of safe communication zones where high-uncertainty physical patches can be observed "on the way"). Through the combination of these cost terms, path search naturally tends to choose a compromise path that is neither too risky nor too conservative.

[0051] like Figure 1 As shown, in step S5, a path optimization search is performed based on the path topology map to generate and execute a replanning path. The replanning path includes: a reflection communication path that uses the target reflection node selected from each candidate reflection node using the path optimization search results as a signal reflection relay, and / or an active detection path aimed at reducing uncertainty parameters in the flight airspace.

[0052] Perform path optimization search (e.g., a graph search-based shortest path algorithm) on the path topology graph. The cost-optimal path from the starting node to the ending node is the replanned path. Among the candidate reflection nodes traversed by the replanned path, the node actually selected as the signal reflection relay is the target reflection node.

[0053] It should be noted that the replanning path is not limited to a single type, but may include one or a combination of two of the following, depending on the specific determination result of the communication threat level in step S3: a reflection communication path and an active detection path. When the communication threat level determination result indicates that the local waiting flight path is about to enter an area where the predicted signal strength is lower than a preset warning threshold, the replanning path includes a reflection communication path—guiding the UAV to fly through the target reflection node to maintain the communication link. When the communication threat level determination result indicates that there is an area of ​​interest ahead of the local waiting flight path where the uncertainty parameter exceeds a preset uncertainty threshold, the replanning path includes an active detection path—guiding the UAV to actively collect radio frequency data of the relevant physical patches of the area of ​​interest to reduce uncertainty. When both conditions are met simultaneously, the replanning path includes both a reflection communication path and an active detection path, which are merged into a coherent flight trajectory.

[0054] The active detection path, aimed at reducing uncertain parameters in the flight airspace, is generated by the following process: identifying physical patches from electromagnetic geometric data that cause the communication threat level of the local waiting flight path to reach a preset threshold as target physical patches; planning a trajectory that deviates from the local waiting flight path and guides the UAV to the observation pose as the active detection path; wherein, the observation pose is a spatial pose that can collect radio frequency data toward the target physical patch, so that the UAV can collect radio frequency data of the target physical patch when executing the active detection path.

[0055] The determination of the observation pose needs to simultaneously meet the following constraints: First, there must be an unobstructed line-of-sight path between the observation pose and the target physical surface, and the angle between the observation direction and the normal vector of the target physical surface must be within a preset effective observation angle range to ensure that the radio frequency data collected at this pose can effectively reflect the reflection contribution of the target physical surface; Second, the spatial location of the observation pose must have a predicted signal strength in the electromagnetic signal prediction field that is higher than a preset safety threshold to ensure that the UAV's communication link is not interrupted during the detection maneuver; Third, the deviation distance between the observation pose and the local flight path must not exceed a preset maximum yaw distance to control the flight cost of the detection maneuver. Among the candidate positions that meet the above constraints, the position whose line-of-sight direction is closest to the normal vector direction of the target physical surface is selected as the observation pose—because when observing close to the normal vector direction, the reflected signal components are most concentrated, and the calibration accuracy of the measured data for the electromagnetic reflection properties of the target physical surface is the highest.

[0056] It is important to note that the design of the active detection path reflects the deep coupling between the path planning process and the electromagnetic sensing process in this scheme. In traditional schemes, the UAV passively receives radio frequency signal data from its current location during flight, and the location from which data is collected depends entirely on the flight path itself. However, in this scheme, the path planning process actively specifies "where to collect what data" based on the distribution of uncertainty parameters—the selection of the target physical patch is driven by uncertainty parameters, the calculation of the observation pose is determined by the spatial geometric position of the target physical patch, and the value of the entire active detection path lies in acquiring measured data for model calibration. As described in the aforementioned reverse calibration process, this measured data will reversely correct the uncertainty parameters in the initial profile information, and the corrected uncertainty parameters will change the judgment result of the communication threat level—forming a closed loop of "planning drives detection, detection corrects the model, and the model changes the planning." Removing the active detection path degenerates the method into static planning that relies solely on offline pre-calculation; removing the distribution information of uncertainty parameters degenerates the active detection path into aimless blind scanning. The two cannot operate independently.

[0057] During the replanning process, continuous reverse calibration of real-time electromagnetic sensing and profile information is performed. Specifically, real-time measured radio frequency data and corresponding real-time pose of the UAV are acquired; electromagnetic signal prediction values ​​corresponding to the real-time pose are extracted from the electromagnetic signal prediction field; the deviation between the measured radio frequency data and the electromagnetic signal prediction values ​​are calculated; and the electromagnetic reflection properties and uncertainty parameters corresponding to the physical patches associated with the electromagnetic geometry data are synchronously corrected based on the deviation.

[0058] This reverse calibration process is continuously performed during each sampling cycle of the UAV's flight. The UAV's RF front-end acquires measured RF data, such as received signal strength and signal-to-noise ratio, in real time at a preset sampling frequency. The visual inertial odometry provides real-time pose estimation independent of the global navigation satellite system. The measured RF data is compared with the predicted value at the corresponding position in the electromagnetic signal prediction field to obtain the deviation value. If the deviation is within a preset normal fluctuation range, the deviation is fused into the electromagnetic signal prediction field using a preset filtering method (such as particle filtering or Kalman filtering). This locally fine-tunes the signal prediction value of the current position and its surrounding voxels, while slightly reducing the uncertainty parameter of the associated physical patch—because the measured data provides a verification of the electromagnetic reflection properties of the physical patch, and the uncertainty of the model's reflection effect on the physical patch is reduced regardless of the deviation direction.

[0059] It should be noted that the determination of "associated physical patches" in the above-mentioned conventional deviation correction is as follows: As described in step S1 above, each voxel in the electromagnetic signal prediction field has its identifier and power percentage of the physical patches that contribute the most to the signal prediction of that voxel recorded during pre-construction. When the UAV collects measured RF data at a certain location and calculates the deviation, the associated physical patches that should be corrected can be determined by querying the pre-stored contributing physical patch identifiers in the corresponding voxel at that location. The deviation correction amount is allocated according to the power percentage of each contributing physical patch—the physical patch with the higher the power percentage is allocated a larger correction amount, because the physical patch has a greater impact on the signal prediction at that location, and the deviation is more likely to be attributed to the deviation of the electromagnetic reflection properties of that physical patch. In a simplified implementation, only the single contributing physical patch with the largest power percentage can be corrected, without correcting the secondary contributing physical patches.

[0060] It's important to note that the aforementioned reverse calibration is a continuous, incremental process, not a one-time batch correction. Each time the UAV flies over a location, a local update is performed on the electromagnetic signal prediction field at that location. This makes the electromagnetic signal prediction field no longer a static map loaded at the start of flight, but a dynamic model that is continuously "polished" by measured data during flight—the accuracy of the electromagnetic signal prediction field in areas already flown by the UAV is higher than in areas not yet flown, and the uncertainty parameters are correspondingly lower. This characteristic directly impacts the rolling forward assessment in step S3: as the UAV continues to fly forward, the local flight path is continuously updated, and the new forward assessment is built upon a more accurate model corrected by measured data, rather than the original offline pre-calculation results.

[0061] During the reverse calibration process, topology-level anomalies are detected and judged simultaneously. Specifically, it is determined whether the deviation indicates that the measured RF data is lower than the predicted electromagnetic signal value and the absolute value of the difference exceeds the preset topology failure threshold. If so, the signal arrival angle ray of the measured RF data is extracted, or a spatial line-of-sight connection between the real-time pose and the communication target source is constructed. The signal arrival angle ray or spatial line-of-sight connection is then spatially geometrically intersected with each physical patch in the electromagnetic geometric data to obtain spatial intersection points. If the spatial intersection point uniquely falls within a specific physical patch, then that specific physical patch is determined to be an anomalous physical patch with a topology-level anomaly.

[0062] It should be noted that topology-level anomaly detection and the aforementioned conventional deviation correction are two different levels of response to the same deviation signal, rather than two independent processing flows. They share the same deviation calculation result, but are routed to different processing paths based on the severity of the deviation: when the deviation is within a preset normal fluctuation range, the aforementioned conventional filtering fine-tuning is performed; when the deviation meets the criteria for topology-level anomaly detection, it is determined that this is not a simple numerical accuracy issue, but rather a structural error in the model—there is a fundamental deviation between the electromagnetic reflection properties of the physical surface and the actual situation.

[0063] The determination of topology-level anomalies requires the simultaneous fulfillment of three conditions: the deviation direction is negative (i.e., the measured RF data is worse than the predicted electromagnetic signal value, indicating that a certain expected signal contributing source did not actually provide the expected signal contribution); the deviation amplitude exceeds the preset topology failure threshold (excluding transient deviations caused by normal fluctuations and random fading); and the deviation can be attributed to a specific physical patch (the specific abnormal physical patch is located by spatial geometric intersection, excluding non-physical patch factors such as overall environmental drift).

[0064] The specific method for spatial geometric intersection is as follows: A signal angle of arrival ray is constructed using the measured RF data (if the UAV's RF front-end supports angle of arrival estimation). Alternatively, when angle of arrival estimation is unavailable, a spatial line-of-sight connection from the real-time pose to the communication target source is constructed as a substitute. This ray or connection is then geometrically intersected with each physical patch in the electromagnetic geometric data in three-dimensional space to obtain spatial intersection points. If a spatial intersection point uniquely falls within a specific physical patch, the deviation can be definitively attributed to that physical patch, classifying it as an anomalous physical patch.

[0065] Typical physical causes of topological anomalies include: the original material of the building facade being replaced by an advertising spray film covering layer after renovation (resulting in the actual reflectivity being much lower than the offline annotation value), temporary construction scaffolding blocking the originally usable reflective surface, or partial renovation of the building exterior wall, etc. In these cases, the deviation between the electromagnetic reflection properties of the offline annotation and the actual situation is structural and cannot be eliminated by numerical fine-tuning.

[0066] In response to the determination of topology-level anomalies, a structural reconstruction of the path topology graph is performed. Specifically, in the initial profile information, abnormal physical patches are marked as in a failed state and removed from each candidate reflection node; the search range of candidate reflection nodes is expanded by increasing the spatial search radius and / or adjusting the preset reflection conditions, and second-order reflection links that meet the preset connectivity conditions and contain at least two consecutive physical patch reflections are extracted; the path topology graph is reconstructed based on each candidate reflection node and second-order reflection link after deleting the abnormal physical patches, and a path optimization search is triggered.

[0067] It's important to note that this step performs a structural reconstruction of the path topology graph, not simply an adjustment of edge weights. Removing an anomalous physical patch from the candidate reflection nodes means that all nodes and connected edges related to that physical patch in the path topology graph are removed—the topology of the path graph changes, not just the cost values ​​of some connected edges. If the deleted node happens to be a necessary relay node in the current replanned path, then that path becomes disconnected, and an alternative path must be found in the reconstructed topology graph.

[0068] To address the failure of critical relay nodes, an extended search is performed during reconstruction: firstly, the spatial search radius is increased to include physical patches previously excluded due to distance; secondly, second-order reflection links, which are not normally prioritized, are included in the path topology. A second-order reflection link refers to a link where a signal originates from the transmitter, is reflected by the first physical patch, reaches the second physical patch, and then is reflected again by the second physical patch before reaching the UAV. Compared to first-order reflection links, second-order reflection links have greater path loss (due to two reflection losses and a longer propagation distance) and are not considered a preferred option under normal circumstances. However, after the failure of a critical first-order reflection node, a second-order reflection link may be the only feasible alternative to maintain communication connectivity.

[0069] On the reconstructed path topology, a path optimization search is re-executed to generate alternative replanning paths. If the search successfully finds an alternative path that meets the minimum communication quality constraints, the task is switched to that path. If no feasible path is found after expanding the search, an emergency degrade strategy is triggered—either a safe landing or a limited climb within regulatory limits to escape the obstructed area.

[0070] It is important to note that there is a two-way coupling between topology reconstruction and the aforementioned active probing. Topology reconstruction is triggered by the anomalous deviation signals accumulated during the reverse calibration process—without continuous reverse calibration and deviation monitoring, topology-level anomalies cannot be detected, and topology reconstruction will never be activated. Conversely, the alternative paths generated after topology reconstruction may contain new candidate reflective nodes to be verified (e.g., newly included distant physical patches in the extended search that have not been previously flight-verified). Active probing will be initiated again for these new nodes—while flying towards the alternative paths, the actual reflection effects of the new reflective nodes will be prioritized to confirm the feasibility of the alternative paths.

[0071] Beyond the entire process of a single flight mission, this method also includes the accumulation of profile information knowledge across mission levels. Specifically, the electromagnetic reflection properties and uncertainty parameters corresponding to the synchronously corrected physical patches are uploaded and updated to a preset global profile information database. For physical patches in the global profile information database that are not covered by measured radio frequency data, their corresponding uncertainty parameters are increased according to a preset attenuation model based on the length of time elapsed since the last update, so as to guide the UAV to generate an active detection path to the physical patch during the next path replanning for the flight airspace.

[0072] After each flight mission, all electromagnetic reflection properties and uncertainty parameters of physical patches updated during the flight through reverse calibration are uploaded to the global profile information database. The global profile information database gathers measured calibration data accumulated from all historical flight missions. The uncertainty parameters of a certain physical patch may have been reduced to an extremely low level through repeated calibrations from multiple flight missions—meaning that the reflection effect of that physical patch is already well understood.

[0073] However, the urban built environment is not static. The electromagnetic reflection characteristics of building facades may change due to renovations, billboard replacements, temporary scaffolding erection, and other reasons. If a physical surface is kept at a low uncertainty level based solely on historical data, path planning will still make decisions based on outdated data when the actual material of that surface has changed—the problem will not be discovered until the drone flies over that surface again and triggers topology-level anomaly detection, by which time communication degradation may have already occurred.

[0074] The preset attenuation model is a mechanism designed to address this issue. For physical patches in the global profile database that have not been covered by measured RF data from any flight missions for an extended period, their uncertainty parameters will automatically increase over time. Even if the uncertainty parameters were low during the last calibration, if no flight missions pass through and re-verify them for a long period afterward, trust in them will gradually be lost, and the uncertainty parameters will slowly rise to a level requiring re-verification. In one implementation, the preset attenuation model uses a linear increment function, meaning the increment of the physical patch's uncertainty parameters is proportional to the length of time elapsed since the last measured update, with the increment rate controlled by a preset time attenuation coefficient. The time attenuation coefficient can be preset differently based on the activity level of the building environment in the area where the physical patch is located—physical patches located in areas with high construction activity (such as urban renewal areas) use a larger time attenuation coefficient, causing their uncertainty parameters to rise more quickly; physical patches located in areas with long-term stable building conditions use a smaller time attenuation coefficient. Furthermore, an upper limit is set for the increase in uncertainty parameters; once this upper limit is reached, further increases cease. This upper limit corresponds to the initial uncertainty level that is completely unverified.

[0075] The closed-loop effect of this mechanism is as follows: when the uncertainty parameters of certain physical patches increase to exceed a preset uncertainty threshold due to long-term lack of observation, the next time a flight mission passes through the area, the communication threat assessment in step S3 will mark the corresponding area of ​​the physical patch as a high-uncertainty area of ​​interest. The exploration reward term in the multi-domain joint cost function of steps S4 and S5 will drive the replanning path to be biased towards this area, generating an active detection path to guide the UAV to the physical patch to collect the latest radio frequency data. Thus, each routine flight mission not only completes its own flight mission, but also performs inspection and updates to the global profile information database "along the way," ensuring that the electromagnetic reflection properties of all physical patches in the database are always in a reliable state that has been periodically verified by field measurements.

[0076] In summary, the UAV path replanning method integrating profile information proposed in this application achieves the following through the complete process of steps S1 to S5:

[0077] First, in complex electromagnetic environments such as urban canyons, by binding the three-dimensional geometric features of building facades with electromagnetic reflection properties into a unified physical patch data structure, and pre-constructing an electromagnetic signal prediction field on this basis, path planning has for the first time gained a systematic understanding of the electromagnetic propagation environment of the flight airspace. This enables the prediction of communication threats and the identification of usable reflection resources during the path planning stage, breaking the traditional technical paradigm of separating spatial physical obstacle avoidance and electromagnetic communication maintenance.

[0078] Second, by introducing uncertainty parameters as an independent judgment dimension in communication threat assessment and setting the constraint relationship between exploration penalty terms and exploration reward terms in the multi-domain joint cost function, this method has the ability to distinguish between three states at the path planning level: "certain security", "certain danger" and "uncertainty". In the "uncertainty" state, it generates the behavior of actively probing the security boundary - this behavior mode is not preset by any single cost term, but is the overall characteristic that emerges after the coupling of multiple cost terms.

[0079] Third, by reverse-calibrating the measured radio frequency data collected during the flight of the UAV to the initial profile information, and distinguishing between numerical deviations and topological anomalies in the reverse calibration, the latter is subjected to structural reconstruction of the path topology map rather than simple parameter fine-tuning. This gives the method the ability to self-heal when faced with structural errors in the model—not only correcting the model parameters, but also reconstructing the solution space of the path.

[0080] Fourth, by uploading and updating brief information across missions and by using a time decay mechanism for uncertainty parameters, the local electromagnetic knowledge accumulated in a single flight mission is transformed into global knowledge. This knowledge then drives subsequent flight missions to perform active inspections and probes of areas that have not been observed for a long time. As the number of flights increases, environmental knowledge gradually evolves, and a synergistic relationship of knowledge accumulation and continuous verification is formed among multiple flight missions.

[0081] Exemplary system:

[0082] Figure 2The illustration shows a UAV path replanning system fused with profile information according to an embodiment of this application, including: an information acquisition module, used to acquire the current pose, target waypoint, initial planned path, and initial profile information of the UAV, the initial profile information including: electromagnetic geometric data containing multiple three-dimensional physical patches and their corresponding electromagnetic reflection attributes, an electromagnetic signal prediction field pre-constructed based on the electromagnetic geometric data, and uncertainty parameters characterizing the reliability of electromagnetic reflection attributes; a path extraction module, used to extract local flight paths located ahead of the current pose in the initial planned path; and a communication threat assessment module, used to map the local flight paths to the electromagnetic signal prediction field to obtain the electromagnetic signal prediction distribution along the local flight paths, and to determine the communication threat assessment based on the electromagnetic signal prediction distribution and the corresponding uncertainty parameters. The system determines the communication threat level of local flight paths; a path topology construction module is used to determine whether the communication threat level is greater than a preset threshold. If so, it extracts physical patches that meet preset reflection conditions from electromagnetic geometry data as candidate reflection nodes, and constructs a path topology graph based on the UAV's current pose, target waypoint, and candidate reflection nodes. The connectivity edge cost of the path topology graph integrates physical space movement cost and electromagnetic link loss cost; a path replanning module is used to perform path optimization search based on the path topology graph to generate and execute replanned paths. The replanned paths include: reflection communication paths that use target reflection nodes selected from candidate reflection nodes using path optimization search results as signal reflection relays, and / or active detection paths aimed at reducing uncertain parameters in the flight airspace.

[0083] Exemplary electronic device:

[0084] Figure 3 A block diagram of an electronic device according to an embodiment of this application is illustrated.

[0085] like Figure 3 As shown, the electronic device includes one or more processors and memory.

[0086] A processor can be a central processing unit (CPU) or other form of processing unit with data processing and / or instruction execution capabilities, and can control other components in an electronic device to perform desired functions.

[0087] The memory may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0088] In one example, the electronic device may also include input devices and output devices, which are interconnected via a bus system and / or other forms of connection mechanism (not shown).

[0089] Of course, for the sake of simplicity, Figure 3 Only some of the components of the electronic device relevant to this application are shown in this illustration; components such as buses, input / output interfaces, etc., are omitted. In addition, the electronic device may include any other suitable components depending on the specific application.

[0090] Exemplary computer-readable medium:

[0091] Embodiments of this application may also be computer-readable storage media storing computer program instructions thereon, which, when executed by a processor, cause the processor to perform the steps described in the "Exemplary Methods" section above according to the various embodiments of this application.

[0092] Computer-readable storage media may take the form of any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may, for example, include, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0093] The basic principles of this application have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this application are merely examples and not limitations, and should not be considered as essential features of each embodiment of this application. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not restrict the application from being implemented using the specific details described above.

[0094] The block diagrams of devices, apparatuses, devices, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0095] It should also be noted that in the apparatus, equipment, and methods of this application, the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered as equivalent solutions of this application.

[0096] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of this application. Therefore, this application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0097] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this application to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A method for replanning the path of a drone that integrates brief documentation information, characterized in that: include: The initial profile information of the UAV is obtained, including its current pose, target waypoint, initial planned path, and flight airspace. The initial profile information includes: electromagnetic geometric data containing multiple three-dimensional physical patches and their corresponding electromagnetic reflection properties, an electromagnetic signal prediction field pre-constructed based on the electromagnetic geometric data, and uncertainty parameters characterizing the reliability of the electromagnetic reflection properties. Extract the local flight path that is ahead of the current pose in the initial planned path; The local flight path is mapped to the electromagnetic signal prediction field to obtain the electromagnetic signal prediction distribution along the local flight path, and the communication threat level of the local flight path is determined based on the electromagnetic signal prediction distribution and the corresponding uncertainty parameter. Determine whether the communication threat level is greater than a preset threshold. If so, extract physical patches that meet the preset reflection conditions from the electromagnetic geometry data as candidate reflection nodes, and construct a path topology graph based on the current pose of the UAV, the target waypoint, and each candidate reflection node. The connectivity edge cost of the path topology graph integrates the physical space movement cost and the electromagnetic link loss cost. Based on the path topology map, a path optimization search is performed to generate and execute a replanning path, which includes: a reflection communication path using a target reflection node selected from each of the candidate reflection nodes as a signal reflection relay using the path optimization search results, and / or an active detection path aimed at reducing the uncertainty parameters in the flight airspace.

2. The UAV path replanning method integrating profile information according to claim 1, characterized in that, The execution of the replanning path also includes: Acquire the real-time measured radio frequency data and corresponding real-time pose of the UAV; Extract the electromagnetic signal prediction value corresponding to the real-time pose in the electromagnetic signal prediction field. Calculate the deviation between the measured radio frequency data and the predicted electromagnetic signal value; The electromagnetic reflection properties corresponding to the physical patch associated with the electromagnetic geometry data and the uncertainty parameters are corrected synchronously based on the deviation.

3. The UAV path replanning method integrating profile information according to claim 2, characterized in that, The execution of the replanning path also includes: Determine whether the deviation indicates that the measured radio frequency data is lower than the predicted value of the electromagnetic signal and the absolute value of the difference exceeds a preset topology failure threshold; If so, extract the signal arrival angle ray from the measured radio frequency data, or construct the spatial line-of-sight connection between the real-time pose and the communication target source; The signal arrival angle ray or the line of sight in space is connected to each physical patch in the electromagnetic geometry data to obtain a spatial intersection point. If the spatial intersection point uniquely falls within a specific physical patch, then the specific physical patch is determined to be an anomalous physical patch with a topological anomaly.

4. The UAV path replanning method integrating profile information according to claim 3, characterized in that, Also includes: In the initial profile information, the abnormal physical patch is marked as a failure state, and the abnormal physical patch is deleted from each of the candidate reflection nodes; By increasing the spatial search radius and / or adjusting the preset reflection conditions to expand the search range of the candidate reflection nodes, second-order reflection links that satisfy the preset connectivity conditions and contain at least two consecutive reflections of the physical surface are extracted. The path topology is reconstructed based on the candidate reflection nodes and the second-order reflection links after deleting the abnormal physical patches, and a path optimization search is triggered.

5. The UAV path replanning method integrating profile information according to claim 2, characterized in that, Also includes: The electromagnetic reflection properties corresponding to the physical patch after synchronous correction, as well as the uncertainty parameters, are uploaded and updated to the preset global profile information database. For physical patches in the global profile information database that are not covered by the measured radio frequency data, the corresponding uncertainty parameters are increased according to a preset attenuation model based on the length of time elapsed since the last update, so as to guide the UAV to generate an active detection path to the physical patch during the next path replanning for the flight airspace.

6. The UAV path replanning method integrating profile information according to claim 1, characterized in that, The physical space mobility cost and electromagnetic link loss cost are calculated using a multi-domain joint cost function, which includes: A physical cost term used to characterize the energy consumption of a drone's geometrical flight or the distance it takes to avoid obstacles; Electromagnetic loss cost term used to characterize the degree of signal attenuation; An exploration penalty term that is positively correlated with the uncertainty parameter distributed along the connected edges; and The exploration reward is negatively correlated with the expected reduction in the uncertainty parameter after the drone flies over the connected edge.

7. The UAV path replanning method integrating profile information according to claim 1, characterized in that, The active detection path aimed at reducing the uncertainty parameters within the flight airspace specifically includes: The physical patches that cause the communication threat level of the local waiting flight path to reach the preset threshold are identified from the electromagnetic geometry data as target physical patches; The planned trajectory deviates from the local waiting flight path and guides the UAV to the observation pose as the active detection path; wherein, the observation pose is a spatial pose that can collect radio frequency data toward the target physical surface, so that the UAV can collect radio frequency data of the target physical surface when executing the active detection path.

8. A drone path replanning system integrating summary information, characterized in that, include: The information acquisition module is used to acquire the current pose, target waypoint, initial planned path and initial profile information of the flight airspace of the UAV. The initial profile information includes: electromagnetic geometric data containing multiple three-dimensional physical patches and their corresponding electromagnetic reflection properties, electromagnetic signal prediction field pre-constructed based on the electromagnetic geometric data, and uncertainty parameters characterizing the reliability of the electromagnetic reflection properties. The route extraction module is used to extract local flight paths that are located in front of the current pose in the initial planned path; The communication threat assessment module is used to map the local flight path to the electromagnetic signal prediction field to obtain the electromagnetic signal prediction distribution along the local flight path, and to determine the communication threat level of the local flight path based on the electromagnetic signal prediction distribution and the corresponding uncertainty parameters. The path topology construction module is used to determine whether the communication threat level is greater than a preset threshold. If so, it extracts physical patches that meet the preset reflection conditions from the electromagnetic geometry data as candidate reflection nodes, and constructs a path topology graph based on the current pose of the UAV, the target waypoint, and each candidate reflection node. The connectivity edge cost of the path topology graph integrates the physical space movement cost and the electromagnetic link loss cost. The path replanning module is used to perform path optimization search based on the path topology map to generate and execute replanned paths. The replanned paths include: a reflection communication path that uses a target reflection node selected from each of the candidate reflection nodes as a signal reflection relay using the path optimization search results, and / or an active detection path aimed at reducing the uncertainty parameters in the flight airspace.

9. An electronic device comprising a memory and a processor, characterized in that: The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions, which, when executed by the processor, implement the steps of the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium having computer-executable instructions stored thereon, characterized in that: When the computer-executable instructions are executed by a processor, they implement the steps of the method as described in any one of claims 1 to 7.

Images