A dynamic adaptive visual mark based method for returning unmanned aerial vehicle to home position

Abstract

This invention discloses a method for UAV repositioning based on dynamic adaptive visual markers. Belonging to the field of UAV visual positioning technology, it obtains high-precision distance and distance change rate between the UAV and a dynamic marker terminal. The dynamic marker terminal adjusts the scale and pattern type of the displayed visual marker pattern in real time based on this information, and employs a smooth transition algorithm to ensure continuous tracking by the onboard camera, keeping the marker imaging size always within a preset optimal recognition window. The UAV calculates its relative pose based on the image and executes repositioning flight control. By adaptively adjusting the pattern and size of the dynamic marker according to the current distance stage of the UAV, it effectively solves the problems of poor scale adaptability, susceptibility to vibration and occlusion interference, and insufficient light adaptability of traditional static markers, achieving continuous and robust autonomous repositioning throughout the entire process, from long-distance guidance to centimeter-level precise landing.

Description

Technical Field

[0001] This invention relates to the field of drone positioning technology, and in particular to a drone positioning method based on dynamic adaptive visual markers. Background Technology

[0002] The low-altitude economy, as a typical representative of future industries and new productive forces, is attracting increasing attention from all sectors. Drones, as an important player in the low-altitude economy, are being used more and more widely. In applications such as logistics delivery, inspection, and security, drones frequently need to perform autonomous homing and precise landing tasks. Currently, the mainstream homing method relies on the Global Navigation Satellite System (GNSS), but in environments such as indoors, urban canyons, under bridges, and forests, satellite signals are severely attenuated or even rejected, leading to unreliable homing. Visual marker-based positioning is an important alternative. The conventional approach involves laying static markers with specific patterns at the landing point, using an onboard camera to capture images and calculate the relative pose to control the landing. However, such static visual signage solutions have prominent problems in practical applications: during the descent of a drone from high altitude to ground, the imaging scale of the signage changes by more than a hundredfold. Fixed-size signs are sparse and undetectable at long distances, and easily exceed the field of view or become blurry at close distances. Even with multiple signage switching or zoom lenses, it is difficult to avoid visual tracking interruptions and pose jumps caused by abrupt pattern changes. At the same time, traditional solutions rely on a single sensor such as a barometer or ultrasonic sensor to obtain altitude information, resulting in high measurement noise and susceptibility to environmental interference, leading to untimely or erroneous signage switching, which seriously affects the continuity of homing. In addition, the recognition rate of static printed signs drops sharply under strong light, shadow, dirt, or partial obstruction, and the airflow under the drone can easily blow away non-fixed signs, causing the positioning reference to drift, which directly endangers landing accuracy and safety.

[0003] More importantly, existing static markers are merely fixed geometric patterns and cannot transmit real-time dynamic information about the landing area to drones, such as ground gust wind speed and direction, and whether the landing pad is occupied by foreign objects. This causes drones to land blindly without environmental awareness, posing a significant safety hazard. In scenarios with alternating light and dark, backlighting, or nighttime, the image quality of ordinary markers by visible light cameras deteriorates severely, easily leading to interruption or failure of the repositioning mission.

[0004] For example, Chinese Patent Application No. 202511105801.9, published on November 28, 2025, discloses a visual positioning-based precision landing system and method for unmanned aerial vehicles (UAVs). The system includes a landing platform for UAV landing; a positioning guidance device comprising multiple independently controllable light-emitting systems arranged in layers, an orientation sensing module for sensing the UAV's position, and a controller for controlling the controllable light-emitting units to generate dynamic visual guidance signals. The orientation sensing module transmits the sensed UAV position information to the controller. A UAV-side module identifies the dynamic visual signals from the positioning guidance device, calculates the spatial coordinates of the landing platform, and generates a landing path from the UAV to the landing platform based on the spatial coordinates. The landing method utilizes this system to perform precise UAV landing. The implementation of this invention effectively solves the signal loss problem of traditional solutions, and allows autonomous positioning and precise landing through visual information even when cargo obstructs the bottom visual or laser sensors.

[0005] The above literature describes a method of illuminating markers in corresponding areas based on their location and dimming or adjusting the brightness of markers in other areas. However, for drones at different altitudes, incomplete acquisition of marker ranges and patterns can affect the accuracy of the entire repositioning process. Furthermore, relying solely on illumination cannot guarantee accurate identification of the actual content represented by the dynamic marker in low-light conditions, thus failing to achieve the intended marking function. Summary of the Invention

[0006] This invention provides a method for drone repositioning based on dynamic adaptive visual markers, which can dynamically adjust the pattern and size of the visual markers according to the current position of the drone, thereby providing accurate markers for drone repositioning under different circumstances.

[0007] To achieve the above objectives, this invention provides a method for drone repositioning based on dynamic adaptive visual markers, which repositions the drone to a repositioning point, where a dynamic marker terminal is set up. The method includes the following steps: S1. Obtaining distance information and drone pitch angle information relative to the dynamic marker terminal. The distance information includes multi-source distance data corresponding to the distance between the drone and the dynamic marker terminal obtained by the drone and the distance between the dynamic marker terminal and the drone obtained by the dynamic marker terminal. The multi-source distance data is fused using a Kalman filter algorithm to obtain a real-time distance estimate and distance change rate between the drone and the dynamic marker terminal. S2. Based on the distance information and distance change rate, the preset focal length of the drone camera, the preset initial dynamic marker size, and the drone pitch angle, adjust the size and pattern type of the displayed visual marker pattern in real time, and keep the imaging size and imaging pattern of the visual marker pattern in the drone camera within a preset range. S3. The drone camera captures the size and image of the visual sign pattern to determine the drone's pose relative to the dynamic sign terminal; the drone performs homing flight control based on the pose.

[0008] The above method obtains the distance between the UAV and the dynamic marker terminal through a ranging device on the UAV, and the distance between the dynamic marker terminal and the UAV is measured by the dynamic marker terminal. Then, using these two multi-source data, a Kalman filter algorithm is used to determine the actual distance between the UAV and the marker terminal. Then, based on the estimated actual distance, the rate of change of distance, camera focal length, and other parameters, the scale and pattern of the dynamic marker on the dynamic marker terminal are determined. This allows the pattern on the dynamic marker terminal to be positioned more clearly within the field of view of the UAV's camera, taking into account the distance between the UAV and the dynamic marker terminal, the rate of change of distance, and the pitch angle. This enables the UAV to accurately determine the position of the dynamic marker, and then adjust the UAV's attitude based on the position of the dynamic marker terminal, allowing the UAV to perform accurate and reliable homing operations.

[0009] Furthermore, step S2 specifically includes: S2.1 Determining the horizontal projection distance of the UAV on the dynamic marker based on the distance information and pitch angle, and then comparing the horizontal projection distance with the preset stage distance to determine the stage in which the UAV is located. If it is in the long-distance stage, proceed to step S2.2; if it is in the medium-distance stage, proceed to step S2.3; if it is in the short-distance stage, proceed to step S2.4. S2.2 When the horizontal projection distance is less than the horizontal projection distance corresponding to the preset long distance and the distance change rate is higher than the preset first threshold, it is determined that it is in the long distance stage. A large area and high brightness coarse positioning mark sequence is displayed on the dynamic marking device. The coarse positioning mark sequence adopts a dynamic coding pattern with high contrast and low frequency features, and is supplemented with a timestamp to achieve frame synchronization. S2.3 When the horizontal projection distance is less than the horizontal projection distance corresponding to the preset medium distance and the distance change rate is close to the preset second threshold, it is determined that it is in the long distance stage. The dynamic sign device displays a composite sign with medium scale and medium frequency characteristics. The composite sign includes a central precise positioning pattern and an outer coding pattern. The central precise positioning pattern is a checkerboard pattern and the outer coding pattern contains compressed and coded environmental information. S2.4 When the horizontal projection distance is less than the preset near distance corresponding to the horizontal projection distance and the distance change rate is less than the preset third threshold, it is determined that it is in the near distance stage. The dynamic marking device displays a high-precision positioning pattern and a dynamic yaw guidance mark. The transformation frequency of the high-precision positioning pattern adopts a preset high frequency mode. The rotation speed of the dynamic yaw guidance mark matches the response frequency of the UAV attitude control ring.

[0010] The above settings determine the drone's distance stage based on the pitch angle, and then display different sizes and frequencies of markers on the dynamic marker device according to different distance stages. For longer distances, thicker icons are used so that the drone can more accurately obtain the marker. For medium distances, a marker that is easier to identify in the middle is used. And for close distances, a direct arrow is used to facilitate a quick response, thereby further ensuring the accuracy of identification.

[0011] Furthermore, the coarse positioning marker sequence includes alternating flashing red and blue concentric rings or preset large-size coded markers; the size of the preset large-size coded markers is determined based on the focal length of the preset camera, the maximum imaging distance, and the distance information between the current UAV and the dynamic marker terminal; the high-precision positioning pattern is a checkerboard pattern, and the horizontal distance of the checkerboard pattern is determined based on the focal length of the preset camera, the medium imaging distance, and the distance information between the current UAV and the dynamic marker terminal; the dynamic yaw guidance marker is an arrow marker that rotates dynamically according to the current yaw error of the UAV, and the rotation angular velocity is matched with the yaw angular velocity error of the UAV.

[0012] The above settings include a preset large-size coded marker in the coarse positioning marker sequence. The size of the preset large-size coded marker is determined by the camera's focal length and maximum image distance to set the maximum size of the marker on the dynamic marker terminal. This ensures that the largest coded marker is more reliably within the camera's pixel distance range and can be reliably observed. In addition, the high-precision positioning pattern is set as a checkerboard, and the lateral distance of the checkerboard is set to the distance information between the current drone and the dynamic marker terminal to form a medium image, which can be more accurately identified. Furthermore, dynamic yaw guidance markers are used for prompting at close range.

[0013] Furthermore, the dynamic sign terminal also includes an infrared light-emitting array arranged circumferentially along the display screen; the method further includes: when the ambient light conditions are lower than a preset threshold, the dynamic sign terminal activates the infrared light-emitting array to present a sign graphic consistent with the visible light visual sign pattern in the infrared band, and the UAV automatically switches to infrared sign recognition to calculate the pose.

[0014] The above settings, by adding an infrared light-emitting array, allow for the formation of an infrared marker pattern when ambient light conditions are below a preset threshold. This pattern enables the determination of the pose, ensuring accurate pose determination even under unreliable ambient light conditions. Furthermore, it also includes: when the dynamic marker terminal detects that the vibration caused by the airflow or wind under the UAV exceeds a threshold, the dynamic marker terminal generates N slightly offset redundant sub-markers in the display screen, and the offset of each redundant sub-marker is randomly generated according to a Gaussian distribution; the UAV recognizes the redundant sub-markers and adjusts its pose accordingly to cancel the vibration noise.

[0015] The above settings, by recognizing airflow at the dynamic marker terminal and using multiple redundant sub-markers, adjust the posture based on the recognized redundant sub-markers to counteract vibration noise, thereby ensuring the drone's stable return to its original position.

[0016] Furthermore, the visual sign pattern displayed by the dynamic sign terminal includes multiple redundant sub-signs with the same code, enabling the UAV to complete positioning by capturing at least one complete sub-sign even in the event of partial occlusion, and the coding redundancy of the redundant sub-signs is not less than 50%.

[0017] The above settings ensure that recognition can be achieved even with partial occlusion.

[0018] Furthermore, the border of the visual logo pattern adopts a dynamic flashing mode with pulsed contrast, and the flashing frequency is synchronized with the camera exposure frequency.

[0019] The above settings, through the flashing mechanism, enable more reliable identification.

[0020] Furthermore, step S2 also includes: S2.1 determining the horizontal projection distance of the UAV on the dynamic marker based on the distance information and pitch angle, and then comparing the horizontal projection distance with the preset stage distance to determine the stage in which the UAV is located. If it is in the long-distance stage, proceed to step S2.2; if it is in the medium-distance stage, proceed to step S2.3; if it is in the short-distance stage, proceed to step S2.4. S2.2 When the horizontal projection distance is less than the horizontal projection distance corresponding to the preset far distance and the distance change rate is higher than the preset first threshold, it is determined that it is in the far distance stage; the maximum distance of the sign displayed on the dynamic sign terminal is determined according to the current distance information, the preset maximum imaging height and the camera focal length; then the ratio is determined according to the preset size of the dynamic sign and the maximum distance of the sign displayed on the dynamic sign terminal; then the spacing value between multiple graphics in the dynamic sign to be displayed is determined according to the product of the ratio and the preset spacing size of the dynamic sign. S2.3 When the horizontal projection distance is less than the horizontal projection distance corresponding to the preset medium distance and the distance change rate is close to the preset second threshold, it is determined that it is in the medium distance stage. The medium distance of the sign displayed on the dynamic sign terminal is determined according to the current distance information, the preset medium imaging height and the camera focal length. Then, the medium ratio is determined according to the preset size of the dynamic mark and the medium distance of the sign displayed on the dynamic sign terminal. Then, the spacing value between multiple graphics in the dynamic mark to be displayed is determined according to the product of the medium ratio and the preset spacing size of the dynamic mark. S2.4 When the horizontal projection distance is less than the preset near distance corresponding to the horizontal projection distance and the distance change rate is less than the preset third threshold, it is determined that it is in the near distance stage. The near distance of the sign displayed on the dynamic sign terminal is determined according to the current distance information, the preset near imaging height and the camera focal length. Then, the near ratio is determined according to the preset size of the dynamic mark and the near distance of the sign displayed on the dynamic sign terminal. Then, the spacing value between multiple graphics in the dynamic mark to be displayed is determined according to the product of the near ratio and the preset spacing size of the dynamic mark.

[0021] The above settings allow for the determination of the size of the marker to be displayed at different distance stages, based on the required imaging height, focal length, and actual distance information. Then, a ratio is determined between the preset dynamic marker size and the size of the marker to be displayed. The marker is then adaptively enlarged according to the ratio, thus adjusting the size of the marker for convenient and accurate identification.

[0022] Furthermore, step S3 includes: the unmanned camera acquiring the visual marker pattern, and the visual marker pattern having a preset corresponding two-dimensional coordinate; converting the two-dimensional coordinates into the three-dimensional coordinates corresponding to the UAV using the PnP algorithm; and then determining the UAV's pose based on the required three-dimensional coordinates and performing homing flight control.

[0023] The above settings achieve positioning control by converting the two-dimensional coordinates of the visual logo to three-dimensional coordinates, thus facilitating control.

[0024] Furthermore, the dynamic sign terminal is also provided with an outer coding pattern, which includes environmental information collected by the environmental sensors integrated into the dynamic sign terminal. The environmental information includes at least one of wind speed, temperature, light intensity, and landing pad occupancy status.

[0025] The above settings, by setting an outer coding pattern on the dynamic marking terminal, enable the identification of environmental information through the outer coding pattern, making it easy for drones to read. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 This is a flowchart of a drone positioning method based on dynamic adaptive visual markers according to the present invention.

[0028] Figure 2 The following are examples of the dynamic marker pattern changes at different distance stages according to the present invention, wherein (a) coarse positioning marker at long distance stage, (b) composite marker at medium distance stage, (c) high-precision positioning pattern at close distance stage, and (d) marker with dynamic yaw guidance.

[0029] Figure 3 This is a diagram of one embodiment of the drone after it has returned to its original position according to the present invention.

[0030] Figure 4 This is a schematic diagram illustrating the calculation of horizontal projection distance in this invention.

[0031] Figure 5 This is a schematic diagram of objects and images in the camera at different stages of the present invention. Detailed Implementation

[0032] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0033] Example 1.

[0034] like Figure 1-3This embodiment provides a method for drone repositioning based on dynamic adaptive visual markers. The drone is repositioned to a designated repositioning point, where a dynamic marker terminal is installed. The dynamic marker terminal includes a display screen, a processor, a first communication module, environmental sensors, a ranging sensor, a three-axis accelerometer, and an infrared emission array. The display screen shows dynamically adaptive visual marker patterns. The processor is responsible for fusing multi-source distance information, calculating distance change rates, generating corresponding visual markers, controlling the output of the display screen and the infrared emission array, and processing environmental and detection data. The first communication module uses a Wi-Fi real-time wireless communication protocol to interact with the drone, ensuring end-to-end communication latency does not exceed 20ms. The environmental sensors collect environmental information such as wind speed, temperature, light intensity, and landing pad occupancy status. The ranging sensor detects the relative distance to the drone. The three-axis accelerometer detects the terminal's vibration frequency and amplitude. The infrared emission array is arranged circumferentially along the display screen to project infrared marker graphics in low-light environments.

[0035] The UAV includes an airborne camera, an infrared camera, an airborne processor, a flight controller, and a second communication module. The airborne camera is used to acquire marker images in the visible light band, and the infrared camera is used to acquire marker images in the infrared band. The airborne processor is responsible for identifying marker features, calculating relative pose, and fusing redundant sub-marker data. The flight controller generates flight control commands based on the pose calculation results and executes homing flight. The second communication module is matched with the first communication module to realize bidirectional data transmission.

[0036] The method includes the following steps: S1. Obtain distance information and pitch angle information of the UAV relative to the dynamic marker terminal. The distance information includes multi-source distance data corresponding to the distance between the UAV and the dynamic marker terminal obtained by the UAV and the distance between the dynamic marker terminal and the UAV obtained by the dynamic marker terminal. The multi-source distance data is fused by Kalman filtering algorithm to obtain the real-time distance estimate and distance change rate between the UAV and the dynamic marker terminal. S2. Based on the real-time distance value and distance change rate, the preset focal length of the drone camera, the preset initial dynamic marker size, and the drone pitch angle, adjust the size and pattern type of the displayed visual marker pattern in real time, and keep the imaging size and imaging pattern of the visual marker pattern in the drone camera within a preset range. S3. The drone camera captures the size and image of the visual sign pattern to determine the drone's pose relative to the dynamic sign terminal; the drone performs homing flight control based on the pose.

[0037] Step S2 specifically includes: S2.1 Determine the horizontal projection distance of the UAV on the dynamic marker based on the real-time distance value and pitch angle, and then determine the stage of the UAV by comparing the horizontal projection distance with the preset stage distance. If it is in the long distance stage, proceed to step S2.2; if it is in the medium distance stage, proceed to step S2.3; if it is in the short distance stage, proceed to step S2.4. S2.2 When the horizontal projection distance is less than the horizontal projection distance corresponding to the preset long distance and the distance change rate is higher than the preset first threshold, it is determined that it is in the long distance stage. A large area and high brightness coarse positioning mark sequence is displayed on the dynamic marking device. The coarse positioning mark sequence adopts a dynamic coding pattern with high contrast and low frequency features, and is supplemented with a timestamp to achieve frame synchronization; then proceed to step S3. S2.3 When the horizontal projection distance is less than the horizontal projection distance corresponding to the preset medium distance and the distance change rate is close to the preset second threshold, it is determined that it is in the medium distance stage. The dynamic sign device displays a composite sign with medium scale and medium frequency characteristics. The composite sign includes a central precise positioning pattern and an outer coding pattern. The central precise positioning pattern is a checkerboard pattern and the outer coding pattern contains compressed and encoded environmental information. Then proceed to step S3. S2.4 When the horizontal projection distance is less than the preset near distance corresponding to the horizontal projection distance and the distance change rate is less than the preset third threshold, it is determined that it is in the near distance stage. The dynamic marking device displays a high-precision positioning pattern and a dynamic yaw guidance mark. The transformation frequency of the high-precision positioning pattern adopts a preset high frequency method. The rotation speed of the dynamic yaw guidance mark matches the response frequency of the UAV attitude control ring. Then proceed to step S3.

[0038] In this embodiment, the specific settings parameters are shown in Table 1 below: Table 1. Relevant parameter values ​​for each stage

[0039] In this embodiment, the coarse positioning marker sequence includes alternating flashing red and blue concentric rings or a preset large-size coded marker; the size of the preset large-size coded marker is determined based on the focal length and maximum imaging distance of the preset camera and the distance information between the current UAV and the dynamic marker terminal; the high-precision positioning pattern is a checkerboard pattern, and the horizontal distance of the checkerboard pattern is determined based on the focal length and medium imaging distance of the preset camera and the distance information between the current UAV and the dynamic marker terminal; the dynamic yaw guidance marker is an arrow marker that rotates dynamically according to the current yaw error of the UAV, and the rotation angular velocity is matched with the yaw angular velocity error of the UAV; the peripheral coded pattern includes environmental information collected by the environmental sensors integrated into the dynamic marker terminal, and the environmental information includes at least one of wind speed, temperature, light intensity, and landing pad occupancy status.

[0040] Step 1: Deploy a dynamic marker terminal at the homing point. After the UAV initiates the homing process, it acquires the distance information between the UAV and the homing point through an onboard LiDAR, ultrasonic sensor, or monocular vision ranging module. Simultaneously, the dynamic marker terminal measures the distance between itself and the UAV using its onboard LiDAR ranging sensor. The data from both ends are exchanged via a real-time wireless communication link. The processor of the dynamic marker terminal uses a Kalman filter algorithm to perform state estimation and noise filtering on the multi-source ranging data, outputting a smooth and robust real-time distance estimate. Based on the distance data of consecutive frames, the distance change rate is calculated. Specifically, the processing of the multi-source ranging data includes: averaging the distance information between the UAV and the homing point measured by the UAV and the distance between the dynamic marker terminal and the UAV measured by the dynamic marker terminal; then inputting this average value into the Kalman filter algorithm to estimate the current distance value and the distance value at the next time step; and then determining the distance change rate by the difference between the distance value at the next time step and the current distance value and the difference between the preset time interval. The Kalman filter algorithm is an existing technology.

[0041] The distance change rate is used to predict the UAV's position in the next moment and adjust the marker scale in advance. It also assesses the stability during the homing phase, preventing frequent marker switching due to distance jumps. This step solves the problems of single-source ranging failure and data jitter through multi-source fusion. Its high-precision distance and change rate output are the core inputs for subsequent adaptive marker control.

[0042] Step 3: Image Acquisition and Pose Calculation. The UAV's onboard camera acquires the size and image of the display screen at a fixed frame rate. The corner points or contour features of the marker are extracted through a visual recognition algorithm. Combined with the actual physical size of the image on the display screen of the dynamic marker terminal, the three-dimensional position coordinates and three-axis attitude angles of the UAV relative to the dynamic marker terminal are calculated. The onboard processor of the homing flight control system transmits the calculated pose data to the flight controller. The flight controller generates control commands based on the deviation between the target homing point and the current pose, adjusts the position and attitude of the UAV, gradually approaches the homing point, and finally completes a precise landing.

[0043] In this embodiment, the visual recognition algorithm is an existing image recognition algorithm. It acquires the image on the dynamic sign terminal, and then compares the acquired image with a preset image to determine the distance information corresponding to the image. The process of comparing the acquired image with the preset image is to determine the preset image corresponding to the acquired image through image similarity. Based on the preset image, the current distance stage can be determined. Based on the distance stage, the depth, extrinsic rotation matrix and translation vector of the camera are determined. Then, based on the intrinsic and extrinsic parameters of the camera, the pose of the UAV relative to the dynamic sign terminal is determined according to the size of the sign in the acquired image.

[0044] In this embodiment, the dynamic sign terminal further includes an infrared light-emitting array arranged around the circumference of the display screen; the method further includes: when the ambient light conditions are lower than a preset threshold, the dynamic sign terminal activates the infrared light-emitting array to present a sign graphic consistent with the visible light visual sign pattern in the infrared band, and the UAV automatically switches to infrared sign recognition to calculate the pose.

[0045] Step S3 also includes a vibration anti-interference mechanism. The dynamic marker terminal has a built-in triaxial accelerometer that detects the vibration frequency and amplitude of the dynamic marker terminal in real time. In this embodiment, the working principle of the triaxial accelerometer in detecting the vibration frequency and amplitude of the dynamic marker terminal is as follows: the triaxial accelerometer converts the acceleration signal generated by the vibration into an electrical signal. When an electrical signal is present, vibration is confirmed and recorded. The electrical signal corresponding to the vibration is then converted into amplitude. When an object vibrates, the sensor senses the change in acceleration in all three axes. By performing a Fourier transform on the collected triaxial acceleration signal, the time domain signal can be converted into a frequency domain signal, thereby clearly obtaining the vibration frequency. When the vibration caused by the downdraft of the UAV or the ambient wind exceeds the threshold (vibration frequency exceeding 10Hz or amplitude exceeding 0.5mm), the dynamic marker terminal generates at least three micro-offsets on the display screen. Redundant sub-markers are generated, and the offsets of each redundant sub-marker are randomly generated according to a Gaussian distribution. The UAV uses the RANSAC random sampling consensus algorithm to remove the pose estimation results of abnormal sub-markers and fuses the pose data of the remaining valid sub-markers to effectively offset the visual positioning noise caused by vibration. In this embodiment, the RANSAC random sampling consensus algorithm includes the following steps: First, the poses of all marker points on the dynamic marker terminal are obtained from the UAV. Then, a minimum sample set is randomly selected from all marker points to instantiate a candidate model. Second, this candidate model is used to test all data points. The marker points that meet the candidate model are formed into a consensus set. Then, the size of the consensus set is evaluated. If it exceeds the historical best record, the best model is updated. Then, the best model is used to determine the redundant marker points that do not meet the requirements. The redundant marker points are removed and the best model is reused to form pose data.

[0046] In this embodiment, the visual marker patterns at each stage are set with multiple redundant sub-markers with the same code by default, and the coding redundancy is not less than 50%. When the marker is partially occluded by birds, debris, etc., the UAV only needs to capture at least one complete sub-marker to complete the positioning calculation, thereby improving the fault tolerance capability in occluded scenarios.

[0047] In this embodiment, the borders of all visual sign patterns adopt a pulsed contrast dynamic flashing mode. The flashing frequency is synchronized with the exposure frequency of the airborne camera, and the pulse duty cycle is adaptively adjusted according to the ambient light intensity: the duty cycle is increased to enhance contrast in strong light environment, and the duty cycle is reduced to avoid overexposure in weak light environment, further improving the sign recognition accuracy under complex lighting conditions.

[0048] The working principle of this invention is as follows: The distance between the UAV and the dynamic marker terminal is obtained through the ranging device on the UAV, and the distance between the dynamic marker terminal and the UAV is measured by the dynamic marker terminal. Then, the actual distance between the UAV and the marker terminal is determined by the Kalman filter algorithm based on these two multi-source data. Then, the scale and pattern of the dynamic marker on the dynamic marker terminal are determined based on the estimated actual distance, the rate of change of distance, the camera focal length, and other parameters. This allows the pattern on the dynamic marker terminal to be positioned more clearly within the field of view of the UAV's camera based on the distance, rate of change of distance, and pitch angle between the UAV and the dynamic marker terminal. This allows for accurate determination of the dynamic marker's position, and the UAV's attitude is adjusted according to the position of the dynamic marker terminal, enabling the UAV to perform accurate and reliable homing operations.

[0049] Example 2.

[0050] The difference between this embodiment and embodiment 1 is that step S2 is implemented differently. In this embodiment, the drone is determined to be in multiple stages by calculation.

[0051] In this embodiment, step S2 includes: S2.1 determining the horizontal projection distance of the UAV on the dynamic marker based on the distance information and the pitch angle, such as... Figure 5 As shown, h represents distance information. Let be the pitch angle of the UAV, where the downward direction of the UAV is set as negative and the upward direction of the UAV is set as positive, and d is the horizontal projection distance of the UAV on the dynamic marker, thus obtaining... The horizontal projection distance of the drone on the dynamic marker was obtained.

[0052] like Figure 5 As shown, the drone is then compared with the preset stage distance to determine the stage it is in. If it is in the long-distance stage, it proceeds to step S2.2; if it is in the medium-distance stage, it proceeds to step S2.3; if it is in the short-distance stage, it proceeds to step S2.4. S2.2 When the horizontal projection distance is less than the preset far distance corresponding to the horizontal projection distance and the distance change rate is higher than the preset first threshold, it is determined that it is in the far distance stage; according to the current distance information h1, the preset farthest imaging height g1 and the camera focal length f, the maximum distance L1 of the sign displayed on the dynamic sign terminal is determined, that is, g1 / L1=(f+f1) / h1, where f1 is the preset distance between the camera image plane 1 and the focal point. Then, the ratio is determined according to the preset size of the dynamic mark and the maximum distance of the sign displayed on the dynamic sign terminal. As the ratio increases, the spacing of the signs displayed on the dynamic sign terminal increases. Then, the spacing value between multiple graphics in the dynamic mark to be displayed is determined according to the product of the ratio and the preset spacing size of the dynamic mark. Then, proceed to step S3. S2.3 When the horizontal projection distance is less than the horizontal projection distance corresponding to the preset medium distance and the distance change rate is close to the preset second threshold, it is determined that it is in the medium distance stage. According to the current distance information h2, the preset medium imaging height g2 and the camera focal length f, the medium distance L2 of the sign displayed on the dynamic sign terminal is determined, that is, g2 / L2=(f+f1) / h2, where f1 is the preset distance between the camera image plane 1 and the focal point. Then, the medium ratio is determined according to the preset size of the dynamic mark and the medium distance of the sign displayed on the dynamic sign terminal. Then, the spacing value between multiple graphics in the dynamic mark to be displayed is determined according to the product of the medium ratio and the preset spacing size of the dynamic mark. Then, proceed to step S3. S2.4 When the horizontal projection distance is less than the preset near distance corresponding to the horizontal projection distance and the distance change rate is less than the preset third threshold, it is determined that it is in the near distance stage. Based on the current distance information g3, the preset near imaging height h3 and the camera focal length f, the near distance L3 of the sign displayed on the dynamic sign terminal is determined, that is, g3 / L3=(f+f1) / h3, where f1 is the preset distance between the camera image plane 1 and the focal point. Then, the near ratio is determined based on the preset size of the dynamic mark and the near distance of the sign displayed on the dynamic sign terminal. Then, the spacing value between multiple graphics in the dynamic mark to be displayed is determined based on the product of the near ratio and the preset spacing size of the dynamic mark. Then, proceed to step S3.

Claims

1. A method for repositioning a drone based on dynamic adaptive visual markers, wherein the drone is repositioned to a designated repositioning point, and a dynamic marker terminal is installed at the repositioning point, characterized in that: Includes the following steps: S1. Obtain distance information and pitch angle information of the UAV relative to the dynamic marker terminal. The distance information includes multi-source distance data corresponding to the distance between the UAV and the dynamic marker terminal obtained by the UAV and the distance between the dynamic marker terminal and the UAV obtained by the dynamic marker terminal. The multi-source distance data is fused by Kalman filtering algorithm to obtain the real-time distance estimate and distance change rate between the UAV and the dynamic marker terminal. S2. Based on the distance information and distance change rate, the preset focal length of the drone camera, the preset initial dynamic marker size, and the drone pitch angle, adjust the size and pattern type of the displayed visual marker pattern in real time, and keep the imaging size and imaging pattern of the visual marker pattern in the drone camera within a preset range. S3. The drone camera captures the size and image of the visual sign pattern to determine the drone's pose relative to the dynamic sign terminal; the drone performs homing flight control based on the pose.

2. The method for UAV relocation based on dynamic adaptive visual markers according to claim 1, characterized in that: Step S2 specifically includes: S2.1 Determine the horizontal projection distance of the UAV on the dynamic marker based on the distance information and pitch angle, and then determine the stage of the UAV by comparing the horizontal projection distance with the preset stage distance. If it is in the long distance stage, proceed to step S2.2; if it is in the medium distance stage, proceed to step S2.3; if it is in the short distance stage, proceed to step S2.

4. S2.2 When the horizontal projection distance is less than the horizontal projection distance corresponding to the preset long distance and the distance change rate is higher than the preset first threshold, it is determined that it is in the long distance stage. A large area and high brightness coarse positioning mark sequence is displayed on the dynamic marking device. The coarse positioning mark sequence adopts a dynamic coding pattern with high contrast and low frequency features, and is supplemented with a timestamp to achieve frame synchronization. S2.3 When the horizontal projection distance is less than the horizontal projection distance corresponding to the preset medium distance and the distance change rate is close to the preset second threshold, it is determined that it is in the long distance stage. The dynamic sign device displays a composite sign with medium scale and medium frequency characteristics. The composite sign includes a central precise positioning pattern and an outer coding pattern. The central precise positioning pattern is a checkerboard pattern and the outer coding pattern contains compressed and coded environmental information. S2.4 When the horizontal projection distance is less than the preset near distance corresponding to the horizontal projection distance and the distance change rate is less than the preset third threshold, it is determined that it is in the near distance stage. The dynamic marking device displays a high-precision positioning pattern and a dynamic yaw guidance mark. The transformation frequency of the high-precision positioning pattern adopts a preset high frequency mode. The rotation speed of the dynamic yaw guidance mark matches the response frequency of the UAV attitude control ring.

3. The method for UAV relocation based on dynamic adaptive visual markers according to claim 2, characterized in that: The coarse positioning mark sequence includes alternating flashing red and blue concentric rings or preset large-size coded marks; The size of the preset large-size coded mark is determined based on the preset camera's focal length, maximum imaging distance, and the distance information between the current drone and the dynamic marker terminal. The high-precision positioning pattern is a checkerboard pattern, and the horizontal distance of the checkerboard pattern is determined based on the preset camera's focal length, medium imaging distance, and the distance information between the current drone and the dynamic marker terminal. The dynamic yaw guidance mark is an arrow mark that rotates dynamically according to the drone's current yaw error, and the rotation angular velocity is matched with the drone's yaw angular velocity error.

4. The method for UAV relocation based on dynamic adaptive visual markers according to claim 1, characterized in that: The dynamic sign terminal also includes an infrared light-emitting array arranged around the circumference of the display screen; the method further includes: when the ambient light conditions are lower than a preset threshold, the dynamic sign terminal activates the infrared light-emitting array to present a sign graphic consistent with the visible light visual sign pattern in the infrared band, and the UAV automatically switches to infrared sign recognition to calculate the pose.

5. The method for UAV relocation based on dynamic adaptive visual markers according to claim 1, characterized in that: Also includes: When the dynamic marker terminal detects that the vibration caused by the airflow or wind under the UAV exceeds the threshold, the dynamic marker terminal generates N slightly offset redundant sub-markers on the display screen. The offset of each redundant sub-marker is randomly generated according to a Gaussian distribution. The UAV recognizes the redundant sub-markers and adjusts its posture accordingly to cancel the vibration noise.

6. The method for UAV relocation based on dynamic adaptive visual markers according to claim 1, characterized in that: The visual sign pattern displayed by the dynamic sign terminal includes multiple redundant sub-signs with the same code, enabling the UAV to complete positioning by capturing at least one complete sub-sign even when partially obscured. The coding redundancy of the redundant sub-signs is not less than 50%.

7. The method for UAV relocation based on dynamic adaptive visual markers according to claim 1, characterized in that: The border of the visual logo pattern adopts a dynamic flashing mode with pulsed contrast, and the flashing frequency is synchronized with the camera's exposure frequency.

8. The method for UAV relocation based on dynamic adaptive visual markers according to claim 1, characterized in that: Step S2 further includes: S2.1 Determining the horizontal projection distance of the UAV on the dynamic marker based on the distance information and pitch angle, and then comparing the horizontal projection distance with the preset stage distance to determine the stage in which the UAV is located. If it is in the long-distance stage, proceed to step S2.2; if it is in the medium-distance stage, proceed to step S2.3; if it is in the short-distance stage, proceed to step S2.

4. S2.2 When the horizontal projection distance is less than the horizontal projection distance corresponding to the preset far distance and the distance change rate is higher than the preset first threshold, it is determined that it is in the far distance stage; the maximum distance of the sign displayed on the dynamic sign terminal is determined according to the current distance information, the preset maximum imaging height and the camera focal length; then the ratio is determined according to the preset size of the dynamic sign and the maximum distance of the sign displayed on the dynamic sign terminal; then the spacing value between multiple graphics in the dynamic sign to be displayed is determined according to the product of the ratio and the preset spacing size of the dynamic sign. S2.3 When the horizontal projection distance is less than the horizontal projection distance corresponding to the preset medium distance and the distance change rate is close to the preset second threshold, it is determined that it is in the medium distance stage. The medium distance of the sign displayed on the dynamic sign terminal is determined according to the current distance information, the preset medium imaging height and the camera focal length. Then, the medium ratio is determined according to the preset size of the dynamic mark and the medium distance of the sign displayed on the dynamic sign terminal. Then, the spacing value between multiple graphics in the dynamic mark to be displayed is determined according to the product of the medium ratio and the preset spacing size of the dynamic mark. S2.4 When the horizontal projection distance is less than the preset near distance corresponding to the horizontal projection distance and the distance change rate is less than the preset third threshold, it is determined that it is in the near distance stage. The distance of the sign displayed on the dynamic sign terminal is determined according to the current distance information, the preset near imaging height and the camera focal length. Then, the near ratio is determined according to the preset size of the dynamic mark and the near distance of the sign displayed on the dynamic sign terminal. Then, the spacing value between multiple graphics in the dynamic mark to be displayed is determined according to the product of the near ratio and the preset spacing size of the dynamic mark.

9. The method for UAV relocation based on dynamic adaptive visual markers according to claim 1, characterized in that: Step S3 includes: the unmanned camera captures the visual marker pattern, and the visual marker pattern has a preset corresponding two-dimensional coordinate. The two-dimensional coordinate is converted into the three-dimensional coordinate corresponding to the UAV through the PnP algorithm. Then, the UAV's pose is determined according to the three-dimensional coordinate required by the UAV and the homing flight control is performed.

10. The method for UAV relocation based on dynamic adaptive visual markers according to claim 1, characterized in that: The dynamic sign terminal is also equipped with an outer coding pattern, which includes environmental information collected by the environmental sensors integrated into the dynamic sign terminal. The environmental information includes at least one of wind speed, temperature, light intensity, and landing pad occupancy status.

Images