Searching method and device used in case of loss of unmanned aerial vehicle

Abstract

The invention discloses a searching method and device used under the condition that an unmanned aerial vehicle is lost, and belongs to the technical field of unmanned aerial vehicle positioning, and the method comprises the steps: obtaining the target position coordinate data of an unmanned aerial vehicle end and the local position coordinate data of a handheld end, and constructing a relative geographic vector model; performing azimuth calculation through the relative geographic vector model to obtain an azimuth angle guidance value; a time domain interaction response value is obtained by establishing a pulse wireless communication distance measurement link and combining an azimuth angle guidance value, and is used for constructing a distance constraint model. According to the invention, through fusion of a relative geographic vector model and an azimuth angle guidance value, with assistance of time domain response of a pulse wireless communication distance measurement link, a distance constraint model is constructed, azimuth-distance coupling analysis is deepened, and a dynamic adjustment mechanism of a search area is derived, so that amplification of coordinate deviation is inhibited in the moving process of the unmanned aerial vehicle, and the accuracy of the movement of the unmanned aerial vehicle is improved. The adaptability of the overall positioning frame is enhanced, and the interference of environmental noise on the guidance value is relieved.

Description

Technical Field

[0001] This invention relates to the field of drone positioning technology, and specifically to a method and apparatus for locating lost drones. Background Technology

[0002] In the field of power line inspection, drones are mainly used to replace manual labor in completing dirty and messy information collection tasks. They reach designated locations by preset coordinate points to collect data such as fault points, and then quickly transmit this information to ground personnel for analysis, judgment and solution formulation, thereby reducing labor intensity and improving inspection efficiency and safety. However, due to the inconsistent quality of drone products, the difference in the skill level of operators, and the control factors in certain airspace, drones often go out of control and crash.

[0003] Currently, there are two main methods for finding crashed drones. One method involves conducting a large-scale search using the last location information recorded before the drone went out of control. This method requires a large amount of manpower, posing a safety hazard to searchers in remote areas, and the search results do not necessarily guarantee finding the drone. The other method involves loading an RTK differential positioning component onto the drone. This RTK differential positioning technology improves positioning accuracy by using differential correction between ground base stations and satellite signals, much like using additional reference points to calibrate GPS errors, thereby achieving higher positioning accuracy. However, this method is costly, poses security risks when communicating with third-party networks, and the heavy device can affect the efficiency of drone flight inspections. Therefore, existing technologies face problems such as insufficient accuracy, long processing time, low security, and poor economic efficiency when dealing with the search for lost drones. Summary of the Invention

[0004] Therefore, the technical problem to be solved by the present invention is that the existing technology faces problems such as insufficient accuracy, long time consumption, low security and poor economy when dealing with the search for lost drones.

[0005] To address the aforementioned technical issues, a method for locating lost drones is proposed. This method includes: acquiring target location coordinate data from the drone and local location coordinate data from a handheld device, and constructing a relative geographic vector model; performing azimuth calculation using the relative geographic vector model to obtain an azimuth angle guidance value; establishing a pulse wireless communication ranging link and combining it with the azimuth angle guidance value to obtain a time-domain interactive response value, which is used to construct a distance constraint model; analyzing the coupling relationship between the azimuth angle guidance value and real-time ranging data using the distance constraint model to obtain an azimuth-distance dual-drive positioning constraint threshold, and performing a first adjustment to the search area; obtaining a dynamic convergence positioning model based on the azimuth-distance dual-drive positioning constraint threshold and the position change during dynamic movement; and locating the drone using the coordinates based on vector intersection obtained through the dynamic convergence positioning model and iteratively updated distance information.

[0006] As a preferred embodiment of the method for locating a lost drone according to the present invention, the following steps are included: acquiring the target location coordinate data of the drone and the local location coordinate data of the handheld device, and constructing a relative geographic vector model: collecting latitude and longitude information of the drone in real time through a satellite positioning unit and transmitting it to the computing control center via a remote wireless transmission protocol; simultaneously collecting the local latitude and longitude information of the handheld device and projecting the target latitude and longitude information onto the local latitude and longitude information; and calculating the latitude difference vector and longitude difference vector based on the projection mapping result to construct a relative geographic vector model.

[0007] As a preferred embodiment of the method for locating lost drones according to the present invention, the method for obtaining azimuth angle guidance values ​​by performing azimuth calculation through the relative geographic vector model includes: extracting the difference between the latitude of the target location and the latitude of the local location as the opposite side vector parameter based on the relative geographic vector model; extracting the difference between the longitude of the target location and the longitude of the local location as the bottom side vector parameter; constructing a trigonometric function relationship through the opposite side vector parameter and the bottom side vector parameter, calculating the tangent value and parsing it to obtain the azimuth angle guidance value.

[0008] As a preferred embodiment of the method for locating lost drones according to the present invention, the following steps are taken: by establishing the pulse wireless communication ranging link and combining it with the azimuth angle guidance value, the time-domain interactive response value is obtained. This includes: monitoring the signal strength threshold of the pulse wireless signal as the drone approaches the target along the azimuth angle guidance value; activating the bilateral two-way ranging protocol when the signal strength meets the preset access conditions and the pulse time interval meets the microsecond-level constraint; triggering pulse signal interaction between master and slave nodes, recording the signal transmission time and signal arrival time during multiple interactions, and generating the time-domain interactive response value.

[0009] As a preferred embodiment of the method for locating lost drones according to the present invention, the following steps are taken: Combining the distance constraint model, analyzing the coupling relationship between the azimuth angle guidance value and real-time ranging data to obtain the azimuth-range dual-drive positioning constraint threshold includes: extracting the first forward transmission time, the first reverse response time, the second forward response time, and the second reverse transmission time from the time-domain interactive response value; processing the time values ​​through cross-product differential operation, and calculating the signal flight time based on the ratio of the sum of the time values; introducing a frequency constant correction factor based on the signal flight time to generate a distance measurement value; using the distance measurement value as a radius constraint, and combining it with the ray constraint of the azimuth angle guidance value, constructing a geometric intersection region to obtain the azimuth-range dual-drive positioning constraint threshold.

[0010] As a preferred embodiment of the method for locating lost drones according to the present invention, the dynamic convergence positioning model is obtained based on the azimuth-range dual-drive positioning constraint threshold and the position change during dynamic movement. This includes: obtaining the frequency operating constant and crystal oscillation error coefficient, and constructing an error compensation function; using the error compensation function to linearly correct the distance measurement value to obtain the corrected distance radius; generating a dynamic circular search boundary on a geographic map based on the corrected distance radius; and updating the intersection area of ​​the circular search boundary and the azimuth ray in real time by combining the displacement increment generated by the handheld device during movement, thereby generating the dynamic convergence positioning model.

[0011] As a preferred embodiment of the method for locating lost drones according to the present invention, the following steps are taken: obtaining coordinates based on vector intersection through the dynamic convergence positioning model and the iteratively updated distance information includes: reducing the chord length range of the search boundary by continuously moving the step size in the dynamic convergence positioning model; calculating the dynamic distance change rate between the current position and the target position in real time; determining a unique vector intersection point based on the convergence trend of the dynamic distance change rate and the chord length range, and outputting the coordinates.

[0012] The beneficial effects of this invention are as follows: by fusing the relative geographic vector model with the azimuth angle guidance value, and supplementing it with the time domain response of the pulse wireless communication ranging link, a distance constraint model is constructed, the azimuth-distance coupling analysis is deepened, and a dynamic adjustment mechanism for the search area is derived. This suppresses the amplification of coordinate deviation during the movement of the UAV, enhances the adaptability of the overall positioning framework, and alleviates the interference of environmental noise on the guidance value.

[0013] By relying on the dual-drive positioning constraint threshold of azimuth and range, and incorporating dynamic position changes, a convergent positioning model is created. Combined with iterative distance information, vector intersection coordinate calculation is achieved, which improves the tracking continuity of UAVs in changing scenarios, avoids the time delay accumulation of traditional ranging methods, amplifies the inclusiveness of the scheme for complex terrain, and weakens the positioning offset caused by signal fluctuations.

[0014] As a preferred embodiment of the precise retrieval device for lost drones described in this invention, it is characterized by comprising a front-end device installed on the drone body and a rear-end device carried by the search personnel; the front-end device, fixed to the drone fuselage, is used to analyze satellite positioning signals in real time and construct a pulse response channel; the rear-end device is used to establish a remote data link and a near-field pulse ranging link with the front-end device, and carries a computing unit for performing azimuth calculation and dynamic convergence positioning model.

[0015] As a preferred embodiment of the precise retrieval device for lost drones described in this invention, the front-end device includes: an embedded processing unit, a first satellite positioning module, a first remote communication module, and a pulse wireless communication tag; the embedded processing unit establishes data connections with the first satellite positioning module, the first remote communication module, and the pulse wireless communication tag respectively through a serial-to-universal serial bus interface; the front-end device includes a metal encapsulation shell, the interior of which is filled with a cured silicone layer; a dual-mode positioning antenna and a pulse communication antenna are fixed to the top outer side of the metal encapsulation shell by mechanical bolts, and a double-sided adhesive layer is provided at the bottom; the front-end device includes a light-emitting diode indicator light, which is electrically connected to the universal input / output interface of the embedded processing unit.

[0016] As a preferred embodiment of the accurate retrieval device for lost drones described in this invention, the back-end device includes a tablet computing device, which is connected to a second satellite positioning module and a pulse wireless communication base station via a data interface.

[0017] The beneficial effects of this invention are as follows: By analyzing satellite signals and constructing pulse channels through the front-end device, combined with the back-end remote link and ranging mechanism, continuous location transmission and high-precision proximity positioning are achieved after the UAV is lost, reducing the blindness of the search and improving the overall retrieval efficiency.

[0018] The backend orientation calculation uses a function to calculate the angle based on the coordinate difference, combined with a dynamic convergence model to gradually reduce the uncertainty area. It integrates multi-source data to avoid single dependence, thereby enhancing the adaptability and positioning reliability in complex environments. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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.

[0020] Figure 1 This is a general flowchart of a method for locating a lost drone, provided as an embodiment of the present invention.

[0021] Figure 2 This is a schematic diagram of the initial positioning of a lost drone based on the fusion of BeiDou and UWB, which is provided as an embodiment of the present invention for a method of finding a lost drone.

[0022] Figure 3 This is a schematic diagram illustrating the gradual reduction of the search area during the dynamic convergence localization process of a method for finding a lost drone, provided in an embodiment of the present invention.

[0023] Figure 4 This is a schematic diagram of trigonometric function calculation for a precise locating device in the event of a lost drone, provided as an embodiment of the present invention.

[0024] Figure 5 This is a schematic diagram of a handheld positioning interface for a precise locating device in the event of a lost drone, provided as an embodiment of the present invention, which displays the drone's location and distance in real time.

[0025] Figure 6 This is a system module structure diagram of a device for accurately locating a lost drone, provided as an embodiment of the present invention. Detailed Implementation

[0026] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0027] Example 1, referring to Figure 1 This is the first embodiment of the present invention, which provides a method for locating a lost drone, including: S1: Obtain the target location coordinate data from the drone and the local location coordinate data from the handheld device, and construct a relative geographic vector model.

[0028] S2: Perform orientation calculations using a relative geographic vector model to obtain orientation angle guidance values.

[0029] S3: By establishing a pulse wireless communication ranging link and combining the azimuth angle guidance value, the time domain interactive response value is obtained and used to construct the distance constraint model.

[0030] S4: Combining the distance constraint model, analyze the coupling relationship between the azimuth angle guidance value and the real-time ranging data, obtain the dual-drive positioning constraint threshold of azimuth and distance, and make the first adjustment of the search area.

[0031] S5: Based on the dual-drive positioning constraint threshold of orientation and distance, a dynamic convergent positioning model is obtained by combining the position change during dynamic movement.

[0032] S6: The UAV is located by using a dynamic convergence positioning model and combining iteratively updated distance information to obtain coordinates based on vector intersection.

[0033] It should be noted that there are currently two main methods for finding crashed drones. The first method is to search a wide area using the last location information of the drone before it went out of control. This method is extremely labor-intensive, and searching in remote areas is not safe for personnel. Moreover, it is not guaranteed that the drone will be found. The second method is to load a real-time dynamic differential positioning component onto the drone. This method is costly, poses security risks when communicating with third-party networks, and the heavy weight of the device affects the drone's patrol efficiency.

[0034] Therefore, to address the aforementioned problems, this invention constructs a precise UAV search method without real-time differential analysis through steps S1 to S6. First, the target coordinates of the UAV and the local coordinates of the handheld device are acquired to establish a relative geographic vector model and clarify the spatial correspondence. Then, based on this model, azimuth calculation is performed to obtain preliminary azimuth guidance. Next, a ranging link is constructed using pulse wireless communication, combining azimuth information with time-domain interactive response to form a distance constraint model, introducing high-precision distance limits. Then, the coupling characteristics of azimuth and real-time ranging are analyzed to obtain a dual-drive azimuth-range positioning threshold, thus narrowing the initial search range. Furthermore, by incorporating position changes during movement, a dynamic convergence positioning model is constructed to correct positioning errors in real time. Finally, vector intersection calculation is performed based on the dynamic convergence model and iterative ranging data to obtain the precise location of the UAV, thereby achieving rapid, decimeter-level search and positioning after a crash.

[0035] Specifically, in step S1: acquire the target location coordinate data from the drone and the local location coordinate data from the handheld device, and construct a relative geographic vector model, including the following steps A1-A3: A1: The latitude and longitude information of the UAV is collected in real time through the satellite positioning unit and sent to the computing control center via a remote wireless transmission protocol; A2: Synchronously collect local latitude and longitude information from the handheld device, and project and map the target latitude and longitude information with the local latitude and longitude information; A3: Based on the projection mapping results, calculate the latitude difference vector and longitude difference vector to construct a relative geographic vector model.

[0036] In this application embodiment, step A2, regarding data collection, is based on local latitude and longitude information sampled at equal intervals, including the following steps A211-A213: A211: If the scenario is a drone search mission in a mountainous area, the crash location of the drone is unknown, and the handheld device is configured as a tablet, the system involved in the positioning calculation is required to maintain data continuity during the approach to the target.

[0037] A212: Real-time monitoring of handheld device location changes. The system receives the output of the local BeiDou module via a serial port and is set to collect latitude and longitude parameters once per second. Each time a sample is collected, the system records the latitude and longitude values ​​at the current moment, i.e., one data point per second, forming a complete local location time series.

[0038] For example, during the search phase from 9:00 AM to 10:00 AM, the system recorded a total of 3,600 sets of latitude and longitude data. These data, arranged chronologically, constitute the local location coordinate data for that phase. This data is a sequence of latitude and longitude data collected at fixed time intervals by the handheld device during its movement, used for subsequent vector model construction and orientation calculation. Each data point represents the actual location of the handheld device at that specific second.

[0039] A213: Local location coordinate data is used as the reference for calculating the relative geographic vector. If the latitude value of a certain second is higher than the latitude reference value of the target location, it is determined that the handheld device is located in the southern region during that time period; otherwise, it is located in the northern region.

[0040] In an optional embodiment, the acquisition method can also be based on smoothed local latitude and longitude information processed by moving average, including the following steps A221-A223: A221: If the scenario is a drone search in a forest area, the complex terrain will cause signal fluctuations. The location data of the handheld terminal may be affected by tree obstruction or multipath effect. Directly using the raw data is prone to jumps.

[0041] A222: A moving average method was used when collecting local latitude and longitude data. Specifically, a current average location value is generated every 10 seconds, which is calculated based on all the raw latitude and longitude data collected in the previous minute.

[0042] For example, at 2 PM, the system will average all latitude and longitude data recorded between 1:59 PM and 2 PM to generate a relatively stable set of representative location values ​​for local reference at the current time. This will be updated again after 10 seconds, and the process will repeat.

[0043] A223: Throughout the entire search phase, the handheld device forms a smooth data sequence consisting of multiple moving average position values, which is the local location coordinate data.

[0044] In another alternative embodiment, for data acquisition, the acquisition method can also be to obtain local location coordinate data based on an event-triggered mechanism, including the following steps A231-A234: A231: Set event trigger conditions. Pre-set trigger rules in the handheld control system or positioning software.

[0045] A232: Real-time monitoring of handheld device movement status, continuous monitoring of location parameters, attention to the changing trend of latitude and longitude curves, and real-time calculation of the rate, direction and amplitude of change to determine whether the triggering conditions are met.

[0046] A233: Once a trigger condition is detected and met, the data acquisition program is immediately started to record the local latitude and longitude data and related information at the current moment.

[0047] A234: Store all event-triggered sampling results in chronological order to form a set of non-equal interval data sequences with movement characteristics, which can be used as local location coordinate data.

[0048] It should be noted that by using the local location coordinate data of the handheld device to construct the vector model, the relative relationship between the handheld device and the target can be accurately reflected, avoiding misjudgment of orientation caused by the deviation between the target location data and the local location data. The collection of real-time or near real-time local data allows the relative geographic vector model to be dynamically updated, which helps search personnel to grasp the location changes in a timely manner and supports path optimization.

[0049] Furthermore, in step S2: azimuth calculation is performed using a relative geographic vector model to obtain the azimuth angle guidance value, including the following steps B1-B3: B1: Based on the relative geographic vector model, extract the difference between the latitude of the target location and the latitude of the local location as the opposite edge vector parameter; B2: Extract the difference between the longitude of the target location and the longitude of the local location as the bottom vector parameter; B3: By constructing a trigonometric function relationship between the side vector parameters and the bottom vector parameters, the tangent value is calculated and the azimuth angle guidance value is obtained analytically.

[0050] In this embodiment of the application, step B1, concerning extraction, is performed by calculating the opposite edge vector parameters based on point-by-point differences, including the following steps B111-B113: B111: If the scenario is a drone positioning task in a plain area, and the handheld device is close to the target within 1 kilometer, the relative geographic vector model contains a continuous sequence of latitude data.

[0051] B112: The system selects the target latitude value and the local latitude value at the current moment from the model, performs a direct subtraction operation, and forms a difference sequence. Each time a calculation is performed, the system records the absolute value and sign of the difference, i.e., a positive difference indicates a northward shift, and a negative difference indicates a southward shift, forming a complete time series of opposite side vectors.

[0052] For example, during the calculation phase from 10:00 AM to 11:00 AM, the system processed a total of 3600 sets of difference data. These data, arranged chronologically, constitute the opposite-edge vector parameters for this phase, which are latitude difference sequences extracted at fixed intervals from the relative geographic vector model, used for subsequent construction of trigonometric function relationships. Each difference represents the vertical offset at that moment.

[0053] B113: The side vector parameter is used as the basis for determining the orientation. If the difference at a certain moment is greater than zero, it is determined that the target is located to the north of the local area during that time period; otherwise, it indicates that the target is located to the south of the local area.

[0054] In an optional embodiment, for the extraction, the extraction method can also be based on smoothed opposite edge vector parameters processed by weighted averaging, including the following steps B121-B123: B121: If the scenario is a drone search in a hilly area, the latitude data may be affected by the terrain undulations, and direct differences are prone to noise.

[0055] B122: A weighted average method was used when extracting the parameters of the opposite side vector. Specifically, a current average difference is generated every 20 seconds. This value is calculated based on all the raw difference data obtained in the previous 2 minutes, with more recent data given higher weight.

[0056] For example, at 3 PM, the system will take a weighted average of all the difference data recorded between 2:58 PM and 3 PM to generate a relatively stable vector representative value for the current location reference. This will be updated again after 20 seconds, and the process will repeat.

[0057] B123: Throughout the entire solution process, a smooth data sequence consisting of multiple weighted average differences is formed, which is the edge vector parameter.

[0058] In another alternative embodiment, for extraction, the extraction method can also be to obtain the opposite edge vector parameters based on a threshold filtering mechanism, including the following steps B131-B134: B131: Set threshold filtering conditions and pre-set filtering rules in the positioning software.

[0059] B132: Real-time monitoring of difference changes, continuous monitoring of latitude differences, attention to the fluctuation range of the difference sequence, and real-time calculation of the degree of deviation to determine whether the filtering conditions are met.

[0060] B133: Once a difference is detected that exceeds the preset threshold, the filtering program is immediately started to adjust or discard the difference and record the corrected vector information.

[0061] B134: Store all filtered results in chronological order to form an optimized data sequence, which is used as the edge vector parameter.

[0062] It should be noted that by using the side vector parameters for azimuth calculation, the relative offset in the vertical direction can be accurately captured, avoiding the interference of latitude difference noise on angle calculation, and extracting real-time or near-real-time differences. This allows the azimuth angle guidance value to be dynamically adjusted, which helps search personnel to correct the direction in a timely manner and supports efficient approach to the target.

[0063] Furthermore, in step S3: by establishing a pulse wireless communication ranging link and combining the azimuth angle guidance value, the time-domain interactive response value is obtained, which is used to construct the distance constraint model, including the following steps C1-C3: C1: The signal strength threshold of the pulsed wireless signal during the process of approaching the target along the azimuth angle guide value; C2: When the signal strength meets the preset access conditions and the pulse time interval meets the microsecond-level constraint, activate the bilateral bidirectional ranging protocol; C3: Triggers pulse signal interaction between master and slave nodes, records the signal transmission and arrival times during multiple interactions, and generates time-domain interaction response values.

[0064] In this embodiment of the application, step C2, regarding activation, is a bilateral two-way ranging protocol triggered based on a signal strength threshold, including the following steps C211-C213: C211: If the scenario is a drone search mission in a suburban area, and the handheld device has entered the target's 600-meter range, the pulse wireless signal begins to be received stably.

[0065] C212: The system monitors the signal strength. When the signal strength exceeds a preset threshold of less than 41 dB and the pulse interval is less than 1 ns, the protocol is immediately activated.

[0066] For example, during the ranging phase from 4 PM to 5 PM, the system executed hundreds of interaction records. These records, arranged sequentially, constitute the activation sequence for this phase, which is the bilateral, bidirectional ranging protocol activated at dynamic intervals during the approach process, used to generate time-domain interaction response values. Each interaction represents the link state at that moment.

[0067] C213: The bilateral, two-way ranging protocol is used as a mechanism to ensure ranging accuracy. If the strength of a certain interaction is higher than a threshold, the link is considered reliable during that period; otherwise, it indicates that we need to wait for the signal to improve.

[0068] In an optional embodiment, for activation, the activation method can also be a progressive bilateral two-way ranging protocol based on time interval constraints, including the following steps C221-C223: C221: If the scenario is a drone search in a dense forest area, the signal may attenuate intermittently, and direct activation is likely to fail.

[0069] C222: A gradual approach was adopted when activating the bilateral two-way ranging protocol. Specifically, the pulse interval was checked every 5 seconds, and the interaction sequence was only formally started when the microsecond-level constraint was met for three consecutive times.

[0070] For example, at 6 PM, the system will check the interval data from the previous 15 seconds. Once it confirms stability, it will generate an activation signal for the current protocol execution. After another 5 seconds, it will check again, and so on.

[0071] C223: Throughout the ranging phase, a stable protocol chain consisting of multiple progressively activated sequences is formed, which is the bilateral bidirectional ranging protocol.

[0072] In another alternative embodiment, for activation, the activation method can also be to obtain a bilateral two-way ranging protocol based on a composite condition mechanism, including the following steps C231-C234: C231: Set composite activation conditions, pre-set multiple rules in the communication module.

[0073] C232: Real-time monitoring of signal parameters, continuous monitoring of intensity and interval, attention to changing trends, and real-time calculation of comprehensive indicators to determine whether the activation conditions are met.

[0074] C233: Once the composite condition is detected to be met, the protocol program is immediately started and the initial interaction information is recorded.

[0075] C234: Store all activation results in chronological order to form an optimized protocol sequence, which is used as a bilateral two-way ranging protocol.

[0076] It should be noted that by using a bilateral two-way ranging protocol to establish a link, the time-domain interactive response value can be accurately captured, avoiding clock drift interference from one-way ranging, activating real-time or near-real-time protocols, enabling the distance constraint model to be dynamically formed, which helps search personnel obtain ranging data in a timely manner and supports area adjustments.

[0077] Furthermore, in step S4: combining the distance constraint model, the coupling relationship between the azimuth angle guidance value and the real-time ranging data is analyzed to obtain the dual-drive positioning constraint threshold of azimuth and distance, and the first adjustment of the search area is performed, including the following steps D1-D4: D1: Extract the first forward transmission time, the first reverse response time, the second forward response time, and the second reverse transmission time from the time-domain interactive response value; D2: The time quantity is processed by cross-product differential operation, and the signal flight time is calculated by combining the ratio of the sum of the time quantities; D3: Based on the signal flight time, a frequency constant correction factor is introduced to generate distance measurement values; D4: Using the distance measurement value as a radius constraint, combined with the ray constraint of the azimuth angle guide value, a geometric intersection region is constructed to obtain the azimuth-distance dual-drive positioning constraint threshold.

[0078] In this embodiment of the application, step D1, concerning extraction, is based on the time required for interactive sequence parsing, and includes the following steps D111-D113: D111: If the scenario is a drone positioning task in a river valley area, the link has been established, and the time domain interaction response value contains multiple interaction records.

[0079] D112: The system parses each interaction's response value into four time parameters: time from backend to frontend to backend, time from frontend response to backend to backend, time from frontend to backend, and time from backend response to frontend. During each parsing process, the system records the values ​​of these parameters, forming a complete time sequence.

[0080] For example, during the adjustment phase from 7 PM to 8 PM, the system extracted thousands of time-related data points. These data points, arranged in the order of interaction, constitute the extraction sequence for this phase. This sequence is a continuous extraction of time-related data points based on the time-domain interaction response values, used for subsequent flight time calculations. Each time-related data point represents the transmission characteristics of that interaction.

[0081] D113: Time values ​​are used as the basis for distance calculation. If a time value is abnormally large, the interaction is considered to be interfered with; otherwise, the transmission is normal.

[0082] In an optional embodiment, for extraction, the extraction method can also be based on the smoothing time amount of the filtering process, including the following steps D121-D123: D121: If the scenario is a drone search in a mountainous area, the response value may contain noise, and direct extraction is prone to errors.

[0083] D122: A filtering method was used when extracting time data. Specifically, a current average time value is generated every 2 seconds, which is calculated based on all the time data parsed within the previous 10 seconds.

[0084] For example, at 1:00 AM, the system filters the data from the previous 10 seconds to generate a stable set of time values ​​for reference in current time calculations. This is then updated again after 2 seconds, and the process repeats.

[0085] D123: During the entire adjustment phase, a smooth sequence consisting of multiple filtering time values ​​is formed, which is the time value.

[0086] In another alternative embodiment, for extraction, the extraction method can also be to obtain the time quantity based on an anomaly detection mechanism, including the following steps D131-D134: D131: Set anomaly detection conditions and pre-set detection rules in the calculation module.

[0087] D132: Real-time monitoring of response values, continuous monitoring of time quantities, attention to numerical distribution, and real-time calculation of deviations to determine whether normal conditions are met.

[0088] D133: Once an anomaly in a certain time quantity is detected, immediately initiate a correction procedure, adjust the value, and record optimization information.

[0089] D134: Store all detection results sequentially to form a reliable data sequence, which can be used as a time measure.

[0090] It should be noted that by using time-based calculations for flight time, the signal propagation delay can be accurately reflected, the interference of clock asynchrony on ranging can be avoided, and real-time or near-real-time response values ​​can be extracted. This allows distance measurements to be dynamically generated, which helps search personnel to narrow down the search area in a timely manner and supports constraint threshold optimization.

[0091] Furthermore, in step S5: based on the dual-drive positioning constraint thresholds of orientation and distance, and combined with the position change during dynamic movement, a dynamic convergent positioning model is obtained, including the following steps E1-E4: E1: Obtain the frequency operating constant and crystal oscillation error coefficient, and construct the error compensation function; E2: Use the error compensation function to linearly correct the distance measurement value to obtain the corrected distance radius; E3: Generate dynamic circular search boundaries on the geographic map based on the corrected distance radius; E4: Combining the displacement increment generated by the handheld device during movement, the intersection area of ​​the circular search boundary and the azimuth ray is updated in real time to generate a dynamic convergence positioning model.

[0092] In this embodiment of the application, step E3, regarding the generation, is based on a dynamic circular search boundary overlaid with an offline map, including the following steps E311-E313: E311: If the scenario is a drone search mission in a wilderness area, the corrected distance radius has been calculated, and the map is a pre-downloaded level 18 offline version.

[0093] E312: The system draws a circular area on the map with the current local location as the center and a radius equal to the corrected distance. Each time the map is updated, the system records the boundary coordinates of the circle, forming a complete sequence of boundaries.

[0094] For example, during the convergence phase from 11 AM to noon, the system generates hundreds of circular boundaries. These boundaries are arranged chronologically, forming the generation sequence for this phase. These are the circular search boundaries dynamically generated on the map based on the corrected distance radius, used for calculating intersection areas. Each boundary represents the search range at that moment.

[0095] E313: Dynamic circular search boundaries are used to define the search area. If the radius of a boundary is smaller than the previous value, the range for that time period is considered to have narrowed; otherwise, the direction of movement needs to be checked.

[0096] In an optional embodiment, the generation method can also be a smooth dynamic circular search boundary based on rasterization processing, including the following steps E321-E323: E321: If the scene involves drone search in complex terrain, the boundaries may need to be refined, as direct generation may easily overlook details.

[0097] E322: A rasterization method was used when generating the dynamic circular search boundary. Specifically, a current raster boundary is generated every 15 seconds. This value is calculated based on all radius data within the previous minute, after being converted into a map raster.

[0098] For example, at 5 PM, the system rasterizes the data from the previous minute, generating a set of smooth boundary representative values ​​for the current model reference. This is repeated 15 seconds later, and the process continues in a loop.

[0099] E323: During the entire convergence phase, a smooth sequence consisting of multiple grid boundaries is formed, which is the dynamic circular search boundary.

[0100] In another alternative embodiment, the generation method can also be to obtain a dynamic circular search boundary based on a boundary optimization mechanism, including the following steps E331-E334: E331: Set boundary optimization conditions and pre-set optimization rules in the map program.

[0101] E332: Monitors radius changes in real time, continuously monitors the boundary, focuses on shape integrity, and calculates coverage in real time to determine whether the optimization conditions are met.

[0102] E333: Once an irregularity at a boundary is detected, the optimization program is immediately started to smooth the curve and record the adjustment information.

[0103] E334: Store all optimized results in chronological order to form a set of efficient boundary sequences, which can be used for dynamic circular search boundaries.

[0104] It should be noted that by using dynamic circular search boundaries for model construction, range changes can be accurately captured, avoiding excessively large areas caused by uncorrected distances, generating real-time or near-real-time boundaries, allowing the dynamic convergence positioning model to gradually shrink, which helps search personnel to focus in a timely manner and supports the determination of the final intersection point.

[0105] Furthermore, in step S6: the UAV is located by obtaining coordinates based on vector intersection through a dynamically converged localization model and iteratively updated distance information, including the following steps F1-F3: F1: In the dynamic convergence localization model, the chord length range of the search boundary is reduced by continuously moving the step size; F2: Calculates the rate of change of dynamic distance between the current position and the target position in real time; F3: Based on the convergence trend of the dynamic distance change rate and the chord length range, determine the unique vector intersection point and output the coordinates.

[0106] In this application embodiment, step F1, regarding the reduction, is based on a chord length range with decreasing step size, and includes the following steps F111-F113: F111: If the scenario is a drone positioning task in a desert area, the model has been dynamically updated, and the movement step size is set to check once every 100 meters.

[0107] F112: The system calculates the chord length after each movement in the model, which is the length of the intersection segment between the circular boundary and the azimuth ray. Each time the model is reduced, the system records the value of the new chord length, forming a complete sequence of chord lengths.

[0108] For example, during the positioning phase from 8 PM to 9 PM, the system processes dozens of chord length reductions. These sequences, arranged in the order of movement, constitute the reduction sequence for that phase, which represents the range of chord lengths reduced in a step-size manner in the dynamic convergent positioning model, used for trend analysis. Each chord length represents the remaining search width after that movement.

[0109] F113: The chord length range is used as an indicator of convergence. If a chord length is less than 2 meters, the time period is considered to be close to the end; otherwise, it indicates that the movement needs to continue.

[0110] In an optional embodiment, the reduction can also be based on an asymptotic chord length range of rate adjustment processing, including the following steps F121-F123: F121: If the scenario is a nighttime drone search, the movement speed is uneven, and directly reducing speed may be too fast.

[0111] F122: A rate adjustment method was used when reducing the string length range. Specifically, a current adjustment string length is generated every 30 seconds, and this value is calculated based on all movement data within the previous 3 minutes.

[0112] For example, at 2:00 AM, the system will adjust the data from the previous 3 minutes to generate a set of progressively increasing chord length values ​​for current location reference. This process will be repeated 30 seconds later, and so on.

[0113] F123: Throughout the entire positioning phase, a progressive sequence consisting of multiple adjustable chord lengths is formed, which is the chord length range.

[0114] In another alternative embodiment, for the reduction, the reduction method can also be to obtain the chord length range based on a threshold monitoring mechanism, including the following steps F131-F134: F131: Set threshold monitoring conditions and pre-set monitoring rules in the model algorithm.

[0115] F132: Real-time monitoring of chord length changes, continuous monitoring of the range, attention to the rate of reduction, and real-time calculation of the remaining width to determine whether the monitoring conditions are met.

[0116] F133: Once a chord length is detected to be close to the threshold, immediately initiate the confirmation procedure, verify the trend, and record the final information.

[0117] F134: Store all monitoring results sequentially to form a converged data sequence, which is used as the chord length range.

[0118] It should be noted that by using the chord length range to determine the intersection point, the convergence process can be accurately tracked, the interference of distance change rate fluctuations on the coordinates can be avoided, the real-time or near-real-time model can be reduced, and the coordinates based on vector intersection can be accurately output, which helps search personnel to lock the location in time and supports the rapid retrieval by drones.

[0119] Example 2, refer to Figure 2-6The second embodiment of the present invention differs from the first embodiment in that: a retrieval device for a lost drone further includes: a front-end device 1 installed on the drone body and a back-end device 2 carried by the search personnel; the front-end device 1 is fixed to the drone body and is used to analyze satellite positioning signals in real time and construct a pulse response channel; the back-end device 2 is used to establish a remote data link and a near-field pulse ranging link with the front-end device 1, and carries a computing unit 3 for performing azimuth calculation and dynamic convergence positioning model.

[0120] In this embodiment, the front-end device 1 is fixed to the UAV fuselage via an adhesive layer. Its function is to analyze satellite positioning signals in real time (i.e., to collect and process satellite signals to obtain latitude and longitude coordinates through the first satellite positioning module 12, such as Beidou module A) and to construct a pulse response channel (i.e., to establish a wireless channel in response to the back-end ranging signal using the pulse wireless communication tag 14, such as UWB pulse wireless communication tag 14, for short-range high-precision ranging). During normal flight, the front-end device 1 continuously collects and stores position data; after the UAV is lost (e.g., crashes), the device continues to operate, powered by a lithium battery, to ensure continuous signal availability.

[0121] The back-end device 2 is used to establish a remote data link with the front-end device 1 (i.e., through a remote communication module, such as a 4G wireless communication module B, to achieve wireless transmission of coordinate data) and a near-field pulse ranging link (i.e., through a pulse wireless communication base station 23, such as a UWB pulse wireless communication micro base station, to perform ranging with an accuracy of ±10cm, and to calculate the flight time using a bilateral bidirectional ranging method to obtain the distance). The back-end device 2 carries the computing unit 3 (such as positioning software pre-installed on a tablet computer), which performs azimuth calculation (i.e., based on the difference in latitude and longitude between the local position and the target position, the azimuth angle is calculated using trigonometric functions such as tan(θ) = opposite side / base side, etc.). Figure 4 As shown in the figure, the opposite sides are the lengths calculated from the latitude difference and the bottom side is the lengths calculated from the longitude difference) and the dynamic convergence positioning model (that is, combining UWB ranging data to gradually narrow the search range, for example, from an initial error range of 50 meters to within 2 meters, and dynamically adjusting the uncertain position area shown in C by the forward distance).

[0122] In one embodiment provided in this application, the front-end device 1 includes an embedded processing unit 11, a first satellite positioning module 12, a first remote communication module 13, and a pulse wireless communication tag 14. The embedded processing unit 11 establishes data connections with the first satellite positioning module 12, the first remote communication module 13, and the pulse wireless communication tag 14 respectively through a serial-to-universal serial bus interface. The front-end device 1 includes a metal encapsulation shell, the interior of which is filled with a cured silicone layer. A dual-mode positioning antenna 15 and a pulse communication antenna 16 are fixed to the top outer side of the metal encapsulation shell by mechanical bolts, and a double-sided adhesive layer is provided at the bottom. The front-end device 1 includes a light-emitting diode indicator light, which is electrically connected to the universal input / output interface of the embedded processing unit 11.

[0123] In this embodiment, the embedded processing unit 11 (such as a Raspberry Pi) establishes data connections with the first satellite positioning module 12, the first remote communication module 13 (used to periodically send coordinate data to the backend), and the pulse wireless communication tag 14 (such as a UWB pulse wireless communication tag 14, used to respond to the backend ranging signal and establish a pulse response channel) via a serial-to-universal serial bus interface. The processing unit is pre-installed with an operating system (such as Linux) and data processing software, responsible for real-time storage of coordinate data, periodic data forwarding, and providing a 5V power supply to the pulse wireless communication tag 14. During operation, the satellite positioning module collects coordinates and transmits them to the processing unit via the interface. The processing unit then formats the coordinates and sends them through the remote communication module, while simultaneously keeping the pulse tag in response mode.

[0124] The front-end device 1 uses a metal enclosure (such as a lightweight aluminum shell) filled with a cured silicone layer (i.e., canned silicone filler, used to cushion impacts and protect the internal circuitry from damage during a fall). The dual-mode positioning antenna 15 and the pulse communication antenna 16 are fixed to the top outer side of the enclosure with mechanical bolts to ensure stable signal reception and transmission; a double-sided adhesive layer (such as 3M double-sided adhesive) is provided at the bottom for secure attachment to the drone body. In addition, the terminal includes a light-emitting diode indicator (such as an LED indicator), electrically connected via a general-purpose input / output interface (such as GPIO) of the embedded processing unit 11, for flashing illumination in low light or at night to facilitate visual-assisted location.

[0125] In one embodiment provided in this application, the back-end device 2 includes a tablet computing device 21, which is connected to a second satellite positioning module 22 and a pulse wireless communication base station 23 via a data interface.

[0126] In this embodiment, the tablet computing device 21 is connected via a data interface to a second satellite positioning module 22 (such as a Beidou module B, used to collect local location coordinates and realize real-time positioning of search personnel) and a pulse wireless communication base station 23 (such as a UWB pulse wireless communication micro base station, used to communicate with the front-end tag, perform ranging with an accuracy of ±10cm, and calculate the distance using AOA and TDOA methods). In addition, the device is also connected to a remote communication module and a dual-mode antenna.

[0127] The tablet computing device 21 is pre-installed with positioning software, including an offline map (downloaded via a MapDownloader server at level 18) and a location calculation and display program. The software runs automatically, receiving local location information (from the second satellite positioning module 22), target location information (from the remote communication module), and high-precision distance information (from the pulse wireless communication base station 23). Then, it performs orientation calculation (calculating the lengths of opposite sides and the base based on the difference in latitude and longitude, and using trigonometric functions to determine the angle) and dynamic convergence (combining ranging data to narrow the search range, gradually reducing the initial coarse orientation to decimeter-level accuracy).

[0128] In summary, when the drone crashes, the built-in Beidou module of the front-end device 1 sends its coordinate information to the back-end in real time. The back-end then pushes the information to the on-site laptop (or the on-site laptop obtains the device's coordinate information by wirelessly accessing the back-end). Simultaneously, the laptop collects the coordinates of the back-end device 2 (i.e., the pilot's position) in real time. Figure 5 .

[0129] As shown in the diagram above, the location of the lost drone can be roughly determined. The pilot can proceed according to the direction. When the search is within 0.6 kilometers (the UWB ranging module can measure a distance of ≤600 meters without obstruction, i.e., the front and back end devices 2 start communicating and measuring distance via UWB), the accurate distance between the front and back end devices 2 can be measured.

[0130] UWB uses a two-sided, two-way ranging method. Backend device 2 actively initiates the first ranging message, and frontend device 1 responds. After 1 second, frontend device 1 actively initiates ranging, and backend device 2 responds, resulting in four time differences: T10, T11, T20, and T21.

[0131] The flight time is calculated using the formula, expressed as follows: T=(T10*T11-T20*T21) / (T10+T11+T20+T21) The formula for the flight time error in bilateral two-way ranging is expressed as follows: E=T(1-(Ka+kb) / 2)Ka The operating frequency constant of front-end device 1 is close to 1, and the operating frequency constant of back-end device 2 is also close to 1. Ka and Kb are mainly related to the crystal oscillators of the UWB at both ends. This design uses a 10ppm crystal (10ppm error), and Ka and Kb may be 0.99999 or 1.00001. Therefore, the measurement error at a distance of 600 meters is ≤10cm.

[0132] Point A is the backend device 2, and point B is the frontend device 1. Assuming the maximum error of BeiDou positioning is 50 meters, then the location range of frontend device 1 is within point C. Figure 2 .

[0133] Searching for supertarget B, such as Figure 2 For every 100 meters advanced, the position range (C) of the front-end device 1 decreases. When the distance is 100 meters, the position range (C) of the front-end device 1 is calculated to be approximately 12 meters.

[0134] Considering that the error of BeiDou is generally no more than 10 meters, this device can achieve a positioning accuracy of about 2 meters.

[0135] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for locating a lost drone, characterized in that: include, Acquire target location coordinate data from the drone and local location coordinate data from the handheld device, and construct a relative geographic vector model; The azimuth calculation is performed using the relative geographic vector model to obtain the azimuth angle guidance value; By establishing a pulse wireless communication ranging link and combining the azimuth angle guidance value, the time domain interactive response value is obtained and used to construct a distance constraint model; By combining the distance constraint model, the coupling relationship between the azimuth angle guidance value and the real-time ranging data is analyzed to obtain the azimuth-distance dual-drive positioning constraint threshold, and the first adjustment of the search area is carried out. Based on the orientation-range dual-drive positioning constraint threshold, a dynamic convergence positioning model is obtained by combining the position change during dynamic movement. The UAV is located by using the dynamic convergence positioning model and the iteratively updated distance information to obtain coordinates based on vector intersection.

2. The method for locating a lost drone according to claim 1, characterized in that: Acquiring the target location coordinate data from the drone and the local location coordinate data from the handheld device, and constructing a relative geographic vector model includes: The latitude and longitude information of the UAV is collected in real time by the satellite positioning unit and sent to the computing control center via a remote wireless transmission protocol. Simultaneously collect local latitude and longitude information from the handheld device, and project and map the target latitude and longitude information with the local latitude and longitude information; Based on the projection mapping results, the latitude difference vector and longitude difference vector are calculated to construct a relative geographic vector model.

3. The method for locating a lost drone according to claim 2, characterized in that: By performing orientation calculations using the relative geographic vector model, the resulting orientation angle guidance values ​​include... Based on the relative geographic vector model, the difference between the latitude of the target location and the latitude of the local location is extracted as the parameter of the opposite edge vector; Extract the difference between the longitude of the target location and the longitude of the local location as the bottom vector parameter; By constructing a trigonometric function relationship between the side vector parameters and the bottom vector parameters, the tangent value is calculated and the azimuth angle guidance value is obtained analytically.

4. A method for locating a lost drone according to claim 3, characterized in that: By establishing the pulse wireless communication ranging link and combining it with the azimuth angle guidance value, the time-domain interactive response value is obtained, including: During the process of approaching the target along the azimuth angle guide value, monitor the signal strength threshold of the pulse wireless signal; When the signal strength meets the preset access conditions and the pulse time interval meets the microsecond-level constraint, the bilateral bidirectional ranging protocol is activated. Trigger pulse signal interaction between master and slave nodes, record the signal transmission time and signal arrival time during multiple interactions, and generate time-domain interaction response values.

5. A method for locating a lost drone according to claim 4, characterized in that: Based on the aforementioned distance constraint model, the coupling relationship between the azimuth angle guidance value and the real-time ranging data is analyzed, resulting in the following azimuth-range dual-drive positioning constraint thresholds: Extract the first forward transmission time, the first reverse response time, the second forward response time, and the second reverse transmission time from the time-domain interactive response value; The time quantities are processed by cross-product differential operation, and the signal flight time is calculated by combining the ratio of the sum of the time quantities. Distance measurements are generated by introducing a frequency constant correction factor based on the signal flight time. By using the distance measurement value as a radius constraint and combining it with the ray constraint of the azimuth angle guide value, a geometric intersection region is constructed to obtain the azimuth-distance dual-drive positioning constraint threshold.

6. A method for locating a lost drone according to claim 5, characterized in that: Based on the aforementioned azimuth-range dual-drive positioning constraint threshold, and combined with the position change during dynamic movement, a dynamic convergent positioning model is obtained, including: Obtain the frequency operating constant and crystal oscillation error coefficient, and construct an error compensation function; The distance measurement value is linearly corrected using an error compensation function to obtain the corrected distance radius; Generate a dynamic circular search boundary on the geographic map based on the corrected distance radius; By combining the displacement increment generated by the handheld device during movement, the intersection area of ​​the circular search boundary and the azimuth ray is updated in real time to generate a dynamic convergence positioning model.

7. A method for locating a lost drone according to claim 6, characterized in that: The dynamic convergence localization model, combined with iteratively updated distance information, yields coordinates based on vector intersection. In the dynamic convergence localization model, the chord length range of the search boundary is reduced by continuously moving the step size; Calculate the rate of change of dynamic distance between the current location and the target location in real time; Based on the convergence trend of the dynamic distance change rate and the chord length range, a unique vector intersection point is determined and its coordinates are output.

8. A precise retrieval device for lost drones, employing a retrieval method for lost drones as described in any one of claims 1 to 7, characterized in that, Includes a front-end device (1) installed on the drone body and a rear-end device (2) carried by the search personnel. The front-end device (1) is fixed to the fuselage of the UAV; The back-end device (2) is used to establish a remote data link and a near-field pulse ranging link with the front-end device (1), and carries a computing unit (3) for performing azimuth calculation and dynamic convergence positioning model.

9. The precise locating device for lost drones according to claim 8, characterized in that: The front-end device (1) includes an embedded processing unit (11), a first satellite positioning module (12), a first remote communication module (13), and a pulse wireless communication tag (14); the embedded processing unit (11) establishes data connections with the first satellite positioning module (12), the first remote communication module (13), and the pulse wireless communication tag (14) respectively through a serial-to-universal serial bus interface; The front-end device (1) also includes a metal encapsulation shell and a light-emitting diode indicator. The metal encapsulation shell is filled with a cured silicone layer. A dual-mode positioning antenna (15) and a pulse communication antenna (16) are fixed to the top outer side of the metal encapsulation shell by mechanical bolts. A double-sided adhesive layer is provided at the bottom. The light-emitting diode indicator is electrically connected to the universal input / output interface of the embedded processing unit (11).

10. The precise locating device for lost drones according to claim 9, characterized in that: The back-end device (2) includes a tablet computing device (21), which is connected to a second satellite positioning module (22) and a pulse wireless communication base station (23) via a data interface.

Images