Unmanned aerial vehicle cluster using visual identification beacons and cluster formation method

Abstract

The invention relates to the field of target feature recognition and unmanned aerial vehicle cluster formation, in particular to an unmanned aerial vehicle cluster using visual recognition beacons, which comprises a plurality of unmanned aerial vehicle sub-clusters. The unmanned aerial vehicle sub-cluster is composed of a central unmanned aerial vehicle and a plurality of auxiliary unmanned aerial vehicles. The auxiliary unmanned aerial vehicles carry visual identification beacons and fly within the visual range of the central unmanned aerial vehicle in a marshalling mode. The central unmanned aerial vehicle uses a visual system to actively detect and identify the visual identification beacons on the auxiliary unmanned aerial vehicles, obtains the position information of the auxiliary unmanned aerial vehicles, completes formation calculation according to task planning, generates a control instruction, and issues the control instruction to the auxiliary unmanned aerial vehicles, thereby achieving the efficient cooperation and agile formation of the cluster. The invention also comprises a method. According to the method, identity recognition, accurate pose calculation and motion state estimation of a large-scale cluster can be realized, rapid and high-agility formation transformation of the cluster is further supported, the method has the advantage of remarkable low detectability, expensive equipment does not need to be additionally arranged, and unification of high performance and low cost is effectively realized.

Description

Technical Field

[0001] This invention relates to the fields of target feature recognition and drone swarm formation, and in particular to a drone swarm and swarm formation method that utilizes visual recognition beacons. Background Technology

[0002] Currently, most drone swarm formation flights rely on the Global Navigation Satellite System (GPS) and terrestrial radio communication control. GPS signals are relatively weak and susceptible to obstructions (such as indoors, under bridges, or in urban canyons) and electromagnetic interference (such as malicious or natural interference), leading to positioning failures or a sharp decline in accuracy, posing a safety risk to dense formations. Terrestrial radio communication control can only provide position coordinates and cannot directly obtain the drones' own orientation information, making it difficult to achieve complex drone swarm control, such as coordinated turning and formation changes. Furthermore, radio communication is easily interfered with, affecting positioning and causing swarm control to fail. Limited by wireless channel capacity, overall scalability is low. Summary of the Invention

[0003] The purpose of this invention is to provide a method for drone swarming and swarm formation using visual recognition beacons, which mainly solves the problems existing in the prior art. By analyzing the near-infrared coded beacons carried by drones, it is possible to realize the identification of large-scale drone swarms, accurate pose (position and orientation) calculation and motion state estimation, thereby supporting fast and highly agile formation changes of the swarm.

[0004] To achieve the above objectives, the technical solution adopted by this invention is to provide a drone swarm utilizing visual recognition beacons, characterized by comprising multiple drone sub-swarms; each drone sub-swarm consists of a central drone and multiple auxiliary drones; the central drone communicates with a ground control center, which plans and controls the flight path; the auxiliary drones carry visual recognition beacons and fly in formation within the visual range of the central drone; the central drone uses a vision system to actively detect and identify the visual recognition beacons on the auxiliary drones, obtain the position information of the auxiliary drones, then completes formation calculations according to the mission plan and generates control commands, and also sends the control commands to the auxiliary drones, thereby achieving efficient collaboration and agile formation of the swarm.

[0005] Furthermore, the vision system used by the central UAV consists of multiple infrared cameras; the infrared cameras are arranged around the central UAV, enabling them to acquire omnidirectional images and are equipped with a global shutter.

[0006] Furthermore, the infrared camera is equipped with a filter module, and the exposure time corresponds to the maximum speed of the auxiliary drone; the filter module is a narrowband filter, allowing light frequencies to pass through that correspond to the light frequencies of the visual recognition beacon, thereby reducing background interference.

[0007] Furthermore, the visual recognition beacon consists of three or more light-emitting diodes and includes attitude coding and identity coding; the attitude coding provides the orientation information of the auxiliary UAV relative to the central UAV; the identity coding is used to identify the unique identity information of the auxiliary UAV.

[0008] Furthermore, the identity code is provided by the flashing frequency of the light-emitting diode; on the same auxiliary drone, the light-emitting diode flashes at the same fundamental frequency; different auxiliary drones use different fundamental frequencies; the central drone reads the fundamental frequency of any of the light-emitting diodes, thus identifying the identity of the auxiliary drone.

[0009] Furthermore, the attitude encoding is provided by an asymmetric distribution pattern composed of multiple light-emitting diodes; simultaneously, one of the light-emitting diodes in the auxiliary UAV serves as a positioning point and flashes at the base frequency, while the remaining light-emitting diodes flash after adding different modulation periods to the base frequency; the central UAV identifies the asymmetric distribution pattern and the modulation period, and calculates the direction information of the auxiliary UAV.

[0010] The present invention also provides a swarm formation method for a drone swarm using the above-described visual recognition beacon, characterized by comprising the steps of:

[0011] Step S100: Group multiple auxiliary drones and place them near a central drone to form a drone sub-cluster, and then group the multiple drone sub-clusters into a drone cluster.

[0012] Step S200: The ground control center sends a mission plan to the central UAV;

[0013] In step S300, the central UAV uses the vision system to identify the auxiliary UAV through the visual recognition beacon;

[0014] In step S400, the central UAV performs formation calculation and generates control commands based on the identified position information of the auxiliary UAVs and the mission planning.

[0015] In step S500, the central UAV sends the control command to the auxiliary UAV while manipulating its own flight path.

[0016] In step S600, the auxiliary UAV adjusts its flight path according to the control command to achieve formation control;

[0017] In step S700, if the task planning is not completed, proceed to step S300.

[0018] Further, in step S300, when the central UAV identifies the subordinate UAV, the process includes the following steps:

[0019] Step S301: The central UAV synchronizes image frames obtained by multiple infrared cameras;

[0020] Step S302: The central UAV acquires multiple consecutive image frames and calculates the background light intensity of the image frames; if the change in background light intensity is less than a threshold, proceed to step S303; otherwise, proceed to step S304.

[0021] Step S303: For multiple image frames, use a differential algorithm to eliminate the background, thereby eliminating continuous background light; proceed to step S305;

[0022] Step S304: Averaging of multiple image frames to improve the signal-to-noise ratio;

[0023] Step S305: After detecting the light spot of the light-emitting diode, continuously track the position and brightness of the light-emitting diode.

[0024] Step S306: Based on the read positions and brightness of all the light-emitting diodes, perform signal analysis to obtain their fundamental frequency and modulation frequency of flickering;

[0025] Step S307: Group the light-emitting diodes from the same auxiliary drone into the same group, obtain the identity information of the auxiliary drone based on the base frequency, and identify the location point of the auxiliary drone based on the modulation frequency;

[0026] Step S308: Based on the positioning point and the other light-emitting diodes in the same group, and combined with the asymmetric distribution pattern of the light-emitting diodes, calculate the six-degree-of-freedom attitude of the auxiliary UAV.

[0027] Further, in step S306, for each tracked light spot, the signal of its brightness changing over time is extracted, and then a fast Fourier analysis is performed to obtain the fundamental frequency; at the same time, the amplitude envelope of the signal is analyzed in the time domain to detect its modulation mode and obtain the modulation frequency.

[0028] Further, in step S307, firstly, the light-emitting diodes with the same fundamental frequency are selected and grouped into the same group; then, among these light-emitting diodes, the unmodulated light spot is identified, which corresponds to the positioning point of the auxiliary drone; finally, the identity information of the auxiliary drone is obtained by looking up a table based on the fundamental frequency.

[0029] In view of the above technical features, compared with the prior art, it has the following significant advantages:

[0030] 1. This invention can acquire key status information such as the unique number, three-dimensional position, orientation, speed, and attitude of each UAV in real time, providing a solid foundation for dynamic formation switching and collaborative control. Even if a few LEDs are completely blocked, as long as the remaining LEDs can be seen flashing for a short time (even within one modulation cycle), the position of the blocked LEDs can be inferred from the modulation pattern.

[0031] 2. This invention utilizes the optical properties of beacons, offering significant advantages in low detectability and enhancing the anti-detection capabilities of the swarm. In particular, by limiting the emission wavelength of the light-emitting diodes to the near-infrared band, the swarm of drones is prevented from being too conspicuous in special environments, such as at night or in dark environments like tunnels, thus avoiding detection due to beacon illumination.

[0032] 3. This invention makes full use of conventional airborne vision sensors (near-infrared cameras) and low-cost LED beacons, without the need to install expensive dedicated equipment such as GPS and IMU, effectively achieving a balance between high performance and low cost. Attached Figure Description

[0033] Figure 1 This is a schematic diagram of the architecture of a drone sub-swarm formation in a preferred embodiment of the present invention that utilizes visual recognition beacons;

[0034] Figure 2 This is a schematic diagram of an auxiliary drone in a preferred embodiment of the drone swarm utilizing visual recognition beacons according to the present invention;

[0035] Figure 3 This is a flowchart of a preferred embodiment of the drone swarm formation method using visual recognition beacons of the present invention.

[0036] In the diagram: 1 - Central UAV, 2 - Auxiliary UAV;

[0037] 11-Infrared camera; 21-First LED; 22-Second LED; 23-Third LED. Detailed Implementation

[0038] The present invention will be further described below with reference to specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be understood that after reading the teachings of this invention, those skilled in the art can make various alterations or modifications to the invention, and these equivalent forms also fall within the scope defined by the appended claims.

[0039] Please see Figure 1 and Figure 2 This invention discloses a method for drone swarming and swarm formation utilizing visual beacons. As shown in the figure, a preferred embodiment divides a drone swarm into multiple drone sub-swarms. Each drone sub-swarm consists of a central drone 1 and multiple auxiliary drones 2. The central drone 1 communicates with a ground control center, which plans and controls the flight path, such as issuing mission plans. The auxiliary drones 2 do not communicate directly with the ground control center; they only communicate with the central drone 1 with which they are grouped. In a drone swarm, the ground control center only needs to communicate with the central drone 1, which actively manages the auxiliary drones 2 in its group. Different central drones 1 do not communicate with each other. In this architecture, the central drone 1, as the command and control core of the sub-swarm, actively detects, identifies, and calculates the encoded information and precise pose of the near-infrared beacons on the tops of all auxiliary drones 2 through its onboard vision system.

[0040] The auxiliary drone 2 carries a visual identification beacon composed of multiple LEDs and flies in formation within the visual range of its corresponding central drone 1. The number of LEDs is three or more, ensuring sufficient scalability for the drone formation. The LEDs continuously flash at a fundamental frequency or a fundamental frequency plus a modulation frequency, transmitting the auxiliary drone 2's own attitude and identity codes. The attitude code provides the auxiliary drone 2's directional information relative to the central drone 1, while the identity code identifies the unique identity of the auxiliary drone 2. Multiple LEDs are arranged in an asymmetrical pattern on the auxiliary drone 2 to provide attitude and identity codes. The asymmetrical distribution avoids ambiguity in a 180° rotation scene. One of these LEDs is selected as a positioning point. The LED at the positioning point flashes at the fundamental frequency, while the remaining LEDs flash with different modulation periods added to the fundamental frequency. Different auxiliary drones 2 have different fundamental frequencies. Therefore, the fundamental frequency constitutes the identity code of one auxiliary drone 2. As long as the brightness of the LEDs is sufficient, their flashing frequency information can still be detected. By identifying the fundamental frequency of the LED flashing, the central UAV 1 can identify all LEDs belonging to the same auxiliary UAV 2 and further calculate the identity of the auxiliary UAV 2. Then, based on the differences in modulation periods (no modulation at the positioning point, different modulation information at other points), the central UAV 1 can identify the installation positions of different LEDs on the auxiliary UAV 2, thereby further calculating the direction information of the auxiliary UAV 2.

[0041] Specifically, in this embodiment, three light-emitting diodes (LEDs) are used, with wavelengths ranging from 800nm ​​to 2000nm, arranged in an asymmetrical planar triangular pattern. The first LED, 21, serves as a positioning point to indicate the direction of the drone's nose; it is installed at the very front of the nose of the auxiliary drone 2. The second LED, 22, and the third LED, 23, are installed on either side of the rear of the auxiliary drone 2, but in an asymmetrical arrangement. Thus, the first LED 21, the second LED 22, and the third LED 23 form a right-angled triangle, with the right-angle vertex located at the nose. This pattern provides clear directionality while avoiding the ambiguity of a 180-degree rotation. Each auxiliary drone 2 is assigned a unique flashing base frequency (FID). The first LED 21 always flashes at the corresponding base frequency FID. The second LED 22, located on the left side of the tail, adds periodicity to the base frequency FID flashing, flashing for the first 0.5 seconds and then turning off for the next 0.5 seconds. The third LED 23, located on the right side of the tail, also adds periodicity to the base frequency FID flashing, turning off for the first 0.5 seconds and then flashing for the next 0.5 seconds. The different modulation periods of the second light-emitting diode 22 and the third light-emitting diode 23 enable the central UAV 1 to quickly distinguish between the two, thereby determining the direction of the auxiliary UAV 2.

[0042] Taking 15 auxiliary drones as an example, their basic frequencies and modulation modes are shown in the table below.

[0043] Affiliated drone ID baseband LED1 mode LED2 mode LED3 mode 01 20Hz 20Hz (continuous) 20Hz (first 0.5sON) 20Hz (0.5s ON after) 02 25Hz 25Hz (continuous) 25Hz (first 0.5sON) 25Hz (0.5s ON after) 03 30Hz 30Hz (continuous) 30Hz (first 0.5sON) 30Hz (0.5s ON after) 04 35Hz 35Hz (continuous) 35Hz (first 0.5sON) 35Hz (0.5s ON after) 05 40Hz 40Hz (continuous) 40Hz (first 0.5sON) 40Hz (0.5s ON after) 06 45Hz 45Hz (continuous) 45Hz (first 0.5sON) 45Hz (after 0.5s ON) 07 50Hz 50Hz (continuous) 50Hz (first 0.5sON) 50Hz (0.5s ON after) 08 55Hz 55Hz (continuous) 55Hz (first 0.5sON) 55Hz (0.5s ON after) 09 60Hz 60Hz (continuous) 60Hz (first 0.5sON) 60Hz (0.5s ON after) 10 65Hz 65Hz (continuous) 65Hz (0.5s ON) 65Hz (0.5s ON after) 11 70Hz 70Hz (continuous) 70Hz (first 0.5sON) 70Hz (0.5s ON after) 12 75Hz 75Hz (continuous) 75Hz (first 0.5sON) 75Hz (0.5s ON after) 13 80Hz 80Hz (continuous) 80Hz (first 0.5sON) 80Hz (0.5s ON after) 14 85Hz 85Hz (continuous) 85Hz (first 0.5sON) 85Hz (0.5s ON after) 15 90Hz 90Hz (continuous) 90Hz (first 0.5sON) 90Hz (0.5s ON after)

[0044] Multiple infrared cameras 11 are installed on the central UAV 1 as a vision system. These cameras actively detect and identify multiple LEDs on the auxiliary UAV 2, which act as visual identification beacons, thereby obtaining the position information of the auxiliary UAV 2. Based on the mission plan, the central UAV calculates the formation and generates control commands, which are then sent to the corresponding auxiliary UAV 2, achieving efficient swarm collaboration and agile formation. Four infrared cameras 11 are installed around the central UAV 1, forming a 360-degree field of view, ensuring omnidirectional image acquisition and avoiding blind spots that could cause UAV formation control failure. To reduce background light interference and accurately capture the visual identification beacons on the fast-moving auxiliary UAV 2, each infrared camera 11 is equipped with a global shutter, with an exposure time corresponding to the maximum speed of the auxiliary UAV 2 (e.g., less than 1 millisecond). The high-speed shutter also helps reduce background light pollution. Narrow-band filter modules are also installed on the infrared cameras 11. The filter module allows the light frequency to correspond to the light frequency of the light-emitting diodes that constitute the visual recognition beacon on the attached UAV 2 (for example, high transmittance within ±20nm of the wavelength of the light-emitting diode), thereby filtering out as much background light as possible and reducing background interference.

[0045] In this embodiment, the infrared cameras 11 on the central UAV 1 capture images at a frame rate of 240fps. Four cameras are installed on the central UAV 1, one in each of the four corresponding directions. All infrared cameras 11 use a global shutter to ensure that the rolling shutter effect does not occur. According to the Nyquist sampling theorem, the highest detectable frequency is 120 Hz.

[0046] Please see Figure 3 The present invention also discloses a swarm formation method for unmanned aerial vehicle (UAV) swarms using the aforementioned visual recognition beacon. A preferred embodiment includes the following steps:

[0047] Step S1: Plan the drone swarm formation.

[0048] First, place multiple auxiliary drones around a central drone, within the visual recognition range of the central drone. Then, group these auxiliary drones and the central drone into the same group in the mission planning, meaning the central drone is responsible for controlling these auxiliary drones. This forms a drone sub-swarm.

[0049] This process is repeated to build multiple drone sub-clusters, which are then grouped into a complete drone cluster.

[0050] Step S2: Send the task plan.

[0051] The ground control center sends mission plans to the central UAV in each UAV sub-swarm. The mission plan includes flight routes and frequency lookup tables. The flight routes include the central UAV's own flight path, as well as the flight paths of each subordinate UAV it controls. The frequency lookup tables contain mapping information between the base frequency and the subordinate UAV's identity, as well as the modulation frequency information of LEDs at different locations on the subordinate UAVs.

[0052] Step S3: The central UAV identifies the attitude of the auxiliary UAV.

[0053] The central UAV uses a vision system and visual recognition beacons installed on the auxiliary UAVs to identify the identity and location information of the auxiliary UAVs.

[0054] Step S31: Acquire image frames of the surrounding environment.

[0055] The raw images obtained by the central UAV from infrared cameras installed around it are stitched together to form a complete image frame.

[0056] Step S32: Evaluate the background light intensity.

[0057] Depending on the rate of change in background light intensity, the central UAV employs different strategies to eliminate its impact. Therefore, the central UAV calculates the background light intensity of each of the recently acquired image frames. If the change in background light intensity is less than a threshold, proceed to step S33; otherwise, proceed to step S34.

[0058] Step S33: Differential background removal.

[0059] When the background light intensity does not change rapidly, a differential algorithm is used to eliminate the background. First, the background frame is identified based on multiple image frames. Then, the differential image is calculated: Differential image = Current image frame - Background frame. In this way, continuous background light (such as sunlight) is subtracted, retaining only the LED signal. Proceed to step S305;

[0060] Step S34: Improve the signal-to-noise ratio of the image on average.

[0061] If the background light changes rapidly, the average of multiple captured image frames is calculated to improve the signal-to-noise ratio.

[0062] Step S35: Based on the existing information, track the LED signal.

[0063] Once the light spot from the LED is detected, the central drone records its position. By combining this with historical data, the position and brightness of the LED can be quickly and continuously tracked.

[0064] Step S36: Perform signal analysis on the light spot of the LED.

[0065] Based on the most recent consecutive image frames, signal analysis is performed according to the position and brightness of all light-emitting diodes, and the fundamental frequency and modulation frequency of their flickering are obtained respectively.

[0066] Specifically, for each tracked light spot, the signal of its brightness changing over time is extracted, and then a Fast Fourier Analysis is performed to obtain its fundamental flicker frequency in the frequency domain. Simultaneously, in the time domain, the amplitude envelope of the light spot signal is analyzed to detect its modulation mode and obtain the modulation frequency.

[0067] Step S37: Assign auxiliary drones.

[0068] All LEDs from the same affiliated drone have the same fundamental flashing frequency, while those from different affiliated drones have different frequencies. Therefore, based on the fundamental flashing frequency of the LEDs, LEDs belonging to the same affiliated drone can be filtered out and grouped together. Then, using the fundamental frequency as an index, a frequency lookup table is consulted to obtain the affiliated drone's identification information. Meanwhile, LEDs used as positioning points have no modulated light spots, so positioning points can be selected from multiple LEDs. Then, other LEDs are further identified based on their modulation frequencies.

[0069] Step S38: Calculate the attitude of the auxiliary UAV.

[0070] Based on the relative positions of the positioning point and other LEDs in the same group, and combined with the known asymmetric distribution pattern of LEDs, the PnP algorithm is used to calculate the six-degree-of-freedom attitude of the auxiliary UAV.

[0071] Step S4: The central UAV generates control commands.

[0072] The central UAV calculates the formation based on the identified identity information of the subordinate UAVs and their location information, according to the mission plan, and generates control commands for each subordinate UAV in the UAV sub-swarm.

[0073] Step S5: The central UAV sends control commands.

[0074] The central UAV maneuvers itself along the planned flight path and sends corresponding control commands to each auxiliary UAV.

[0075] Step S6: The auxiliary drone executes control commands.

[0076] The auxiliary drones receive control commands, adjust their flight paths, and achieve formation control.

[0077] Step S7: If mission planning is not completed, proceed to step S3 for the next flight control operation. Otherwise, proceed to step S8.

[0078] Step S8, task completed.

[0079] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A drone swarm utilizing visual recognition beacons, characterized in that, It comprises multiple drone sub-clusters; each drone sub-cluster consists of a central drone and multiple auxiliary drones; the central drone communicates with a ground control center, which plans and controls the flight path; the auxiliary drones are equipped with visual recognition beacons and fly in formation within the visual range of the central drone; the central drone uses a vision system to actively detect and identify the visual recognition beacons on the auxiliary drones, obtain the position information of the auxiliary drones, and then completes formation calculations and generates control commands according to the mission plan, and also sends the control commands to the auxiliary drones, thereby achieving efficient collaboration and agile formation of the cluster.

2. The UAV swarm utilizing visual recognition beacons according to claim 1, characterized in that, The vision system used by the central UAV consists of multiple infrared cameras; the infrared cameras are arranged around the central UAV, enabling them to acquire omnidirectional images and are equipped with a global shutter.

3. The UAV swarm utilizing visual recognition beacons according to claim 2, characterized in that, The infrared camera is equipped with a filter module, and the exposure time corresponds to the maximum speed of the auxiliary UAV; the filter module is a narrowband filter, which allows light frequencies to pass through that correspond to the light frequencies of the visual recognition beacon, thereby reducing background interference.

4. The UAV swarm utilizing visual recognition beacons according to claim 1, characterized in that, The visual recognition beacon consists of three or more light-emitting diodes and includes attitude coding and identity coding; the attitude coding provides the orientation information of the auxiliary UAV relative to the central UAV; the identity coding is used to identify the unique identity information of the auxiliary UAV.

5. The UAV swarm utilizing visual recognition beacons according to claim 4, characterized in that, The identity code is provided by the flashing frequency of the light-emitting diode; on the same auxiliary drone, the light-emitting diode flashes at the same fundamental frequency; different auxiliary drones use different fundamental frequencies; the central drone reads the fundamental frequency of any of the light-emitting diodes, thus identifying the identity of the auxiliary drone.

6. The UAV swarm utilizing visual recognition beacons according to claim 5, characterized in that, The attitude encoding is provided by an asymmetric distribution pattern composed of multiple light-emitting diodes; at the same time, one of the light-emitting diodes in the auxiliary UAV serves as a positioning point and flashes at the base frequency, while the remaining light-emitting diodes flash after adding different modulation periods to the base frequency; The central UAV identifies the asymmetric distribution pattern and the modulation period, and calculates the direction information of the auxiliary UAV.

7. A swarm formation method for unmanned aerial vehicle (UAV) swarms using visual recognition beacons as described in claim 1, characterized in that, Includes steps, Step S100: Group multiple auxiliary drones and place them near a central drone to form a drone sub-cluster, and then group the multiple drone sub-clusters into a drone cluster. Step S200: The ground control center sends a mission plan to the central UAV; In step S300, the central UAV uses the vision system to identify the auxiliary UAV through the visual recognition beacon; In step S400, the central UAV performs formation calculation and generates control commands based on the identified position information of the auxiliary UAVs and the mission planning. In step S500, the central UAV sends the control command to the auxiliary UAV while manipulating its own flight path. In step S600, the auxiliary UAV adjusts its flight path according to the control command to achieve formation control; In step S700, if the task planning is not completed, proceed to step S300.

8. The swarm formation method for UAV swarms using visual recognition beacons according to claim 7, characterized in that, In step S300, when the central UAV identifies the auxiliary UAV, the following steps are included: Step S301: The central UAV synchronizes image frames obtained by multiple infrared cameras; Step S302: The central UAV acquires multiple consecutive image frames and calculates the background light intensity of the image frames; If the change in background light intensity is less than the threshold, proceed to step S303; otherwise, proceed to step S304. Step S303: For multiple image frames, use a differential algorithm to eliminate the background, thereby eliminating continuous background light; proceed to step S305; Step S304: Averaging of multiple image frames to improve the signal-to-noise ratio; Step S305: After detecting the light spot of the light-emitting diode, continuously track the position and brightness of the light-emitting diode. Step S306: Based on the read positions and brightness of all the light-emitting diodes, perform signal analysis to obtain their fundamental frequency and modulation frequency of flickering; Step S307: Group the light-emitting diodes from the same auxiliary drone into the same group, obtain the identity information of the auxiliary drone based on the base frequency, and identify the location point of the auxiliary drone based on the modulation frequency; Step S308: Based on the positioning point and the other light-emitting diodes in the same group, and combined with the asymmetric distribution pattern of the light-emitting diodes, calculate the six-degree-of-freedom attitude of the auxiliary UAV.

9. The swarm formation method for UAV swarms using visual recognition beacons according to claim 8, characterized in that, In step S306, for each tracked light spot, the signal of its brightness changing over time is extracted, and then a fast Fourier analysis is performed to obtain the fundamental frequency; at the same time, the amplitude envelope of the signal is analyzed in the time domain to detect its modulation mode and obtain the modulation frequency.

10. The swarm formation method for UAV swarms utilizing visual recognition beacons according to claim 8, characterized in that, In step S307, firstly, LEDs with the same fundamental frequency are selected and grouped into the same group; then, among these LEDs, unmodulated light spots are identified, which correspond to the positioning points of the auxiliary UAV; finally, the identity information of the auxiliary UAV is obtained by looking up a table based on the fundamental frequency.

Images