Method for homeward flight landing and obstacle avoidance of Beidou patrol inspection unmanned aerial vehicle

Abstract

The invention discloses a method for homeward flight landing and obstacle avoidance of a Beidou patrol inspection unmanned aerial vehicle, and relates to the technical field of obstacle avoidance of unmanned aerial vehicles. In the landing optimization stage, the obstacle coverage area, the ground flatness and the electromagnetic interference intensity in the landing area are comprehensively considered, the suitable degree of the landing area is comprehensively evaluated by presetting the allowable value and calculating the landing evaluation index in a weighted mode, the optimal landing position is selected for the unmanned aerial vehicle, the landing risk is reduced, and the landing efficiency is improved. The problems that in the prior art, when the unmanned aerial vehicle reaches the landing position and prepares to land, the suitability degree of the landing area cannot be comprehensively evaluated, the optimal landing position cannot be selected for the unmanned aerial vehicle, the landing risk cannot be reduced, and the intelligent degree is low are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of UAV obstacle avoidance, and specifically to a method for obstacle avoidance during the return landing of a Beidou inspection and patrol UAV. Background Art

[0002] In the current era of booming technology development, UAVs, with their unique advantages, have been widely used in many fields. Whether it is topographic surveying and mapping, agricultural plant protection operations, or film shooting, logistics distribution and other scenarios, UAVs play an important role, bringing great convenience to people's production and life.

[0003] In the return landing link after the UAV completes the task, obstacle avoidance technology becomes the key to ensuring its safe recovery, which is directly related to the integrity rate of the UAV, the integrity of task execution, and the cost-effectiveness of related operations. However, the existing UAV return landing obstacle avoidance methods still have the following deficiencies in actual application:

[0004] When the UAV arrives at the landing position and is about to land, it is unable to comprehensively evaluate the suitability of the available landing areas, select the best landing position for the UAV, reduce the landing risk, and has a low degree of intelligence;

[0005] In addition, for static obstacles, obstacle avoidance can be achieved through advance path planning or real-time detection. However, for dynamic obstacles, such as flying birds, other UAVs, etc., the existing technology still has deficiencies in evaluating the risk and potential hazards of dynamic obstacles. The movement of dynamic obstacles is uncertain, which makes it difficult for the UAV to accurately judge the potential impact of dynamic obstacles on the UAV, and there are limitations.

[0006] Therefore, a method for obstacle avoidance during the return landing of a Beidou inspection and patrol UAV is introduced. Summary of the Invention

[0007] The purpose of the present invention is to solve the problems pointed out in the background art, and to propose a method for obstacle avoidance during the return landing of a Beidou inspection and patrol UAV.

[0008] The purpose of the present invention can be achieved through the following technical solutions: A method for obstacle avoidance during the return landing of a Beidou inspection and patrol UAV, including:

[0009] Landing optimization: When the UAV reaches the position range of the landing point, it reduces the flight height and enters the landing stage. A landing pre-preparation duration is preset during the landing stage. During the landing pre-preparation duration, the landing evaluation index sera of each available landing area within the landing point position range is analyzed, and the final landing position of the UAV is determined;

[0010] Specifically:

[0011] M1: Extract the coverage area of obstacles within each group of landing areas, calculate the proportion of the obstacle coverage area in the total area of the landing areas to obtain the obstacle impact value kz1 for each group of landing areas; obtain the height variance of the point cloud data within each group of landing areas as the ground impact value kz2 for each group of landing areas;

[0012] Use an electromagnetic sensor to detect the electromagnetic interference intensity around each group of landing areas as the electromagnetic interference value kz3 for each group of landing areas;

[0013] M2: Preset the allowable obstacle value, allowable ground value, and allowable electromagnetic value corresponding to the obstacle impact value kz1, ground impact value kz2, and electromagnetic interference value kz3 respectively, denoted as ks1, ks2, and ks3;

[0014] M3: Conduct a comprehensive analysis of the obstacle impact value kz1, ground impact value kz2, and electromagnetic interference value kz3 for each group of landing areas to obtain the landing evaluation index sera for each group of landing areas;

[0015] M4: Select the landing area with the lowest landing evaluation index sera as the final landing position of the drone.

[0016] As a preferred embodiment of the present invention, the specific analysis for obtaining the landing evaluation index sera for each group of landing areas is as follows:

[0017] According to the formula Perform a weighted calculation on the obstacle impact value kz1, ground impact value kz2, and electromagnetic interference value kz3 for each group of landing areas to obtain the landing evaluation index sera for each group of landing areas; where c1, c2, and c3 are the impact weight factors of the obstacle impact value kz1, ground impact value kz2, and electromagnetic interference value kz3 respectively.

[0018] As a preferred embodiment of the present invention, it further includes:

[0019] Path planning: According to the current position of the drone and the position of the landing point, combined with map data and obstacle information, use the high-precision positioning function of the Beidou positioning system to plan the preliminary return landing path corresponding to the drone;

[0020] Real-time obstacle detection: During the flight of the drone along the preliminary return landing path, detect dynamic obstacles within the set distance range of the drone in real time. After a corresponding dynamic obstacle appears within the set distance range of the drone at a certain time point, first mark the time point of appearance as the starting point. After identifying the type of the dynamic obstacle, analyze the movement trajectory of the dynamic obstacle within the set time window after the starting point to determine the potential hazard index σ of the current dynamic obstacle to the flight process of the drone;

[0021] Path optimization: Based on the potential hazard index σ of the current dynamic obstacle to the flight process of the UAV, it is judged whether it is necessary to re-adjust the preliminary return landing path.

[0022] As a preferred embodiment of the present invention, analyze the movement track of the dynamic obstacle within a set time window after the starting point, specifically:

[0023] Extract the three-dimensional coordinates of the dynamic obstacle at each time point within the set time window and form discrete trajectory points. For the three-dimensional coordinates of the UAV's position and the trajectory points of the dynamic obstacle at the same time point, plot the corresponding coordinate points in three-dimensional space. After the plotting is completed, connect the two sets of coordinate points as the hypotenuse of a right triangle; starting from the three-dimensional coordinates of the UAV's position, construct a perpendicular line to the horizontal line where the trajectory point is located as a right side of the right triangle, connect the hypotenuse and the right side to construct a right triangle, and use the right triangles constructed at each time point as the risk models between the dynamic obstacle and the UAV at each time point within the set time window;

[0024] Extract the areas of the risk models at each time point within the set time window and arrange them in chronological order. After the arrangement is completed, calculate the difference in area between adjacent time points; if the difference at a certain time point is greater than 0, it is marked as a risk increase value, and if the difference at a certain time point is less than 0, it is marked as a risk decrease value after taking the absolute value;

[0025] After summing all the risk increase values and risk decrease values respectively, obtain the total increase value and the total decrease value. Subtract the total decrease value from the total increase value to obtain the trend judgment value. If the trend judgment value is positive, it is determined as a risk increase value, and vice versa, it is determined as a risk decrease value after taking the absolute value;

[0026] Preset the risk intervals corresponding to each group of risks for risk increase and risk decrease respectively. Each risk interval for risk increase corresponds to an increase additional coefficient, and each risk interval for risk decrease corresponds to a decrease additional coefficient;

[0027] Based on the determination result of the trend judgment value calculated within the set time window, convert it into an increase additional coefficient or a decrease additional coefficient, denoted as Gi, where i = 1 or 2.

[0028] As a preferred embodiment of the present invention, analyzing the movement track of the dynamic obstacle within a set time window after the starting point further includes:

[0029] After identifying the type of the dynamic obstacle, preset a set of stability additional coefficients and the maximum allowable risk model area corresponding to different types of dynamic obstacles, denoted as λ and yu;

[0030] Calculate the mean value of the risk model area at each time point within the set time window to obtain the risk mean value, and extract the area of the risk model at the end time point among each time point within the set time window as the risk valuation;

[0031] Mark the risk mean value and the risk valuation between the dynamic obstacle and the UAV within the set time window as yt1 and yt2 respectively.

[0032] As a preferred embodiment of the present invention, determine the potential hazard index σ of the current dynamic obstacle to the UAV flight process, specifically:

[0033] According to the formula Perform weighted calculation on the risk mean value and the risk valuation between the dynamic obstacle and the UAV within the set time window to obtain the potential hazard index σ of the current dynamic obstacle to the UAV flight process; where η1 and η2 are the influence weight factors of the risk mean value and the risk valuation respectively.

[0034] As a preferred embodiment of the present invention, based on the potential hazard index σ of the current dynamic obstacle to the UAV flight process, determine whether it is necessary to re-adjust the preliminary return landing path, specifically:

[0035] Identify the type of the current dynamic obstacle, set the threshold index range of the potential hazard index σ corresponding to different types of dynamic obstacles, match the potential hazard index σ of the current dynamic obstacle to the UAV flight process with the threshold index range. If the potential hazard index σ is lower than the threshold index range, push the risk description "The obstacle is far and stable" to the management personnel.

[0036] As a preferred embodiment of the present invention, determining whether it is necessary to re-adjust the preliminary return landing path further includes:

[0037] If the potential hazard index σ is within the threshold index range, push the risk description "The obstacle is approaching but controllable" to the management personnel, and at the same time use the RRT algorithm to generate three candidate paths and push them to the management personnel. After the management personnel receive and replace the corresponding candidate path, adjust the preliminary return landing path;

[0038] If the potential hazard index σ is higher than the threshold index range, push the risk description "The risk impact of the obstacle is high" to the management personnel, and after performing the candidate path push in S2, activate the preset UAV emergency obstacle avoidance setting.

[0039] Compared with the prior art, the beneficial effects of the present invention are:

[0040] 1. In the landing optimization stage of the present invention, by comprehensively considering the obstacle coverage area, ground flatness, and electromagnetic interference intensity within the available landing area, and presetting allowable values and calculating the landing evaluation index through weighted calculation, the suitability of the available landing area is comprehensively evaluated, the best landing position is selected for the UAV, and the landing risk is reduced, solving the problem in the prior art that when the UAV reaches the landing position and is ready to land, the suitability of the available landing area cannot be comprehensively evaluated, the best landing position cannot be selected for the UAV, the landing risk cannot be reduced, and the degree of intelligence is relatively low;

[0041] 2. By constructing a risk model based on a right triangle, the present invention quantifies the spatial relationship between the UAV and dynamic obstacles into an area index, calculates the area of the risk model at each time point within a set time window, and judges whether the risk rises or falls according to the area difference between adjacent time points to obtain a trend judgment value. Based on the trend judgment value, it is converted into a value-added additional coefficient or a value-reducing additional coefficient. At the same time, combined with the stability additional coefficient preset for different types of dynamic obstacles and the maximum allowable area of the risk model, the risk mean value and risk valuation are calculated through weighted calculation to obtain a potential hazard index, solving the problem in the prior art that there are still deficiencies in evaluating the risk hazard degree of dynamic obstacles, and the movement of dynamic obstacles is uncertain, which makes it difficult for the UAV to accurately judge the potential hazard impact of dynamic obstacles on the UAV;

[0042] 3. By using lidar to accurately measure the distance and azimuth between the obstacle and the UAV, combining with the image information provided by the camera, identifying the type and characteristics of the obstacle, scanning the contour of the obstacle in real time through point cloud data, constructing a static environment model and comparing the difference between the current frame and the background, extracting moving targets, and then using the Euclidean clustering algorithm to separate independent moving objects, the three-dimensional coordinates of dynamic obstacles can be accurately obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] For the convenience of those skilled in the art to understand, the present invention will be further described below with reference to the accompanying drawings.

[0044] Figure 1 It is a flowchart of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0045] The technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments of the present invention belong to the scope of protection of the present invention.

[0046] Please refer to Figure 1 As shown, a method for a Beidou inspection and patrol UAV to return and land to avoid obstacles includes:

[0047] Return instruction reception: When the drone receives the return instruction sent from the ground, it obtains its current location through the Beidou positioning system; specifically, the longitude, latitude, and altitude; meanwhile, it switches the flight mode of the drone to the return mode and activates sensors such as lidar and cameras;

[0048] Path planning: Based on the current location of the drone and the landing point location, combined with map data and obstacle information, using the high-precision positioning function of the Beidou positioning system, it plans the preliminary return landing path corresponding to the drone; path planning takes into account the flight performance and safety requirements of the drone, avoids the path passing through known obstacle areas, and at the same time, sets a certain safety buffer zone for the path to cope with possible emergencies;

[0049] Real-time obstacle detection: During the flight of the drone along the preliminary return landing path, it uses sensors such as lidar and cameras to detect dynamic obstacles within the set distance range of the drone in real time. After a dynamic obstacle appears within the set distance range of the drone at a certain time point, first mark the time point as the starting point. After identifying the type of the dynamic obstacle, analyze the movement track of the dynamic obstacle within the set time window after the starting point to determine the potential hazard index σ of the current dynamic obstacle to the flight process of the drone;

[0050] The types of dynamic obstacles include but are not limited to flying birds, other drones, etc.; lidar can accurately measure the distance and azimuth between the obstacle and the drone, while the camera can provide image information of the obstacle for identifying the type and characteristics of the obstacle;

[0051] Due to possible errors and noises in sensor measurements, it is necessary to filter the collected data; common filtering methods include Kalman filtering, mean filtering, etc. to improve the reliability and stability of the data;

[0052] Specifically:

[0053] It scans the obstacle contour within the set distance range of the drone in real time through point cloud data, constructs a static environment model through historical point cloud data, compares the difference between the current frame and the background in real time, extracts moving targets, applies DBSCAN or Euclidean clustering algorithm to the point cloud data to separate independent moving objects, and identifies the type of dynamic obstacles through RGB images;

[0054] Extract the three-dimensional coordinates of a dynamic obstacle at each time point within a set time window and form discrete trajectory points. For the three-dimensional coordinates of the UAV's location and the trajectory points of the dynamic obstacle at the same time point, plot the corresponding coordinate points in three-dimensional space. After the plotting is completed, connect the two sets of coordinate points as the hypotenuse of a right triangle; the length represents the straight-line distance between the UAV and the dynamic obstacle at this time point. Starting from the three-dimensional coordinates of the UAV's location, construct a perpendicular line to the horizontal line where the trajectory point is located as one of the right sides of the right triangle, connect the hypotenuse and the right side to construct a right triangle, and use the right triangles constructed at each time point as the risk model between the dynamic obstacle and the UAV at each time point within the set time window.

[0055] It should be noted that to determine the position and length of the perpendicular line, relevant knowledge of three-dimensional space geometry needs to be used. For example, the vector between the UAV and the dynamic obstacle can be obtained first, and then according to the properties and geometric relationships of the vector, a vector perpendicular to this vector and passing through the UAV coordinate point can be found, so as to determine the direction and position of the perpendicular line.

[0056] Extract the areas of the risk models at each time point within the set time window and arrange them in chronological order. After the arrangement is completed, calculate the difference in area between adjacent time points, that is, subtract the area of the time point on the right from the area of the time point on the left at adjacent time points. If the difference at a certain time point is greater than 0, it is marked as a risk increase value; if the difference at a certain time point is less than 0, it is marked as a risk decrease value after taking the absolute value.

[0057] After summing all the risk increase values and risk decrease values respectively, obtain the total increase value and the total decrease value. Subtract the total decrease value from the total increase value to get the trend judgment value. If the trend judgment value is positive, it is determined as a risk increase value; otherwise, it is determined as a risk decrease value after taking the absolute value.

[0058] Preset the risk intervals corresponding to each group of risks for risk increase and risk decrease values respectively. Each risk interval for risk increase corresponds to an increase additional coefficient, and each risk interval for risk decrease corresponds to a decrease additional coefficient.

[0059] Among them, the range of the increase additional coefficient is set between 1.038 - 1.129, and the range of the decrease additional coefficient is set between 0.839 - 0.947. The greater the risk increase, the higher the possibility of corresponding to 1.129, and the greater the risk decrease, the higher the possibility of corresponding to 0.839.

[0060] Based on the determination result of the trend judgment value calculated within the set time window, convert it into an increase additional coefficient or a decrease additional coefficient, denoted as Gi, where i = 1 or 2, G1 represents the increase additional coefficient, and G2 represents the decrease additional coefficient.

[0061] After identifying the type of dynamic obstacles, a set of stability additional coefficients and maximum areas of the allowed risk model are preset for different types of dynamic obstacles, respectively, and are denoted as λ and yu;

[0062] The stability additional coefficient is set according to the specific type of dynamic obstacles, and the range is set between 1.057-1.138. For example, birds are more variable and less stable, so the stability additional coefficient value should be set higher than that of drones.

[0063] The risk model area at each time point in the set time window is averaged to obtain the risk mean, and the area of ​​the risk model at the end time point in the set time window is extracted as the risk estimate;

[0064] The risk mean and risk estimate between the dynamic obstacle and the UAV within the set time window are marked as yt1 and yt2 respectively;

[0065] According to the formula The risk mean and risk valuation between dynamic obstacles and UAVs within the set time window are weighted and calculated to obtain the potential hidden danger index σ of the current dynamic obstacles to the UAV flight process; where η1 and η2 are the influence weight factors of the risk mean and risk valuation, respectively, and η2>η1;

[0066] It should be noted that by constructing a risk model based on a right triangle, the spatial relationship between the drone and the dynamic obstacle is quantified as an area index, and the risk is assessed based on the change in the risk model area at each time point in the time window. This method can more intuitively and accurately reflect the potential threat level of dynamic obstacles to drone flight, which is more scientific and reasonable than the traditional simple distance judgment or single factor assessment.

[0067] Calculate the area difference between adjacent time points, sum the risk increase value and risk decrease value, and obtain the trend judgment value. Then convert it into an additional coefficient for value increase or value decrease according to the trend judgment value. At the same time, set the stability additional coefficient based on the type of dynamic obstacle, and perform weighted calculation on the risk mean and risk valuation to obtain the potential hidden danger index. This risk assessment method that comprehensively considers multiple factors can more flexibly adapt to different types of dynamic obstacles and changing environments, and improves the accuracy and reliability of risk assessment.

[0068] Different stability additional coefficients and maximum areas of risk models are preset for different types of dynamic obstacles (such as birds in flight, other drones, etc.). For example, for birds with strong variability and poor stability, a higher stability additional coefficient is set, which makes it more cautious when assessing their potential hazards to drones, thereby improving the safety and adaptability of drones when facing different types of obstacles.

[0069] Suppose a drone is flying along a preliminary return landing path. At a certain moment, let's call it t0, a flying bird is detected as a dynamic obstacle. Mark t0 as the starting point and set a time window of 5 seconds, recording relevant data every 1 second.

[0070] Using sensors such as lidar and cameras, obtain the three-dimensional coordinates of the bird (dynamic obstacle) at each time point from t0 to t0 + 5 seconds (i.e., t1 = t0 + 1 second, t2 = t0 + 2 seconds, t3 = t0 + 3 seconds, t4 = t0 + 4 seconds, t5 = t0 + 5 seconds) to form discrete trajectory points O1, O2, O3, O4, O5. At the same time, obtain the three-dimensional coordinates of the drone at the same time points U1, U2, U3, U4, U5.

[0071] For the t1 moment, connect U1 and O1 as the hypotenuse of a right triangle. Its length c1 represents the straight-line distance between the drone and the bird at this time.

[0072] Taking U1 as the starting point, construct a perpendicular line to the horizontal line where O1 is located as one of the right sides. Use three-dimensional space geometry knowledge to calculate the length a1 of this right side, and then obtain the right triangle at the t1 moment, whose area is marked as S1. Similarly, calculate the areas S2, S3, S4, S5 of the right triangles at the t2, t3, t4, t5 moments respectively.

[0073] Calculate the difference in area between adjacent time points: ΔS1 = S1 - S2. If ΔS1 > 0, mark it as a risk increase value; if ΔS1 < 0, take the absolute value and mark it as a risk decrease value.

[0074] Calculate ΔS2 = S2 - S3, ΔS3 = S3 - S4, ΔS4 = S4 - S5 in turn and mark them according to the above rules.

[0075] Suppose ΔS1 = 2 (risk increase value), ΔS2 = -1 (1 after taking the absolute value, risk decrease value), ΔS3 = 3 (risk increase value), ΔS4 = -2 (2 after taking the absolute value, risk decrease value).

[0076] Sum up all the risk increase values: total increase value, and sum up all the risk decrease values: total decrease value.

[0077] Trend judgment value = total increase value - total decrease value. Since the trend judgment value is positive, it is determined as a risk increase.

[0078] Given that the range of the increase additional coefficient is 1.038 - 1.129, because the risk increase is 2, assume that according to the preset rules and corresponding relationships, the increase additional coefficient G1 = 1.08 is determined.

[0079] Since the detected dynamic obstacle is a bird, according to the preset, the value range of the stability additional coefficient of the bird is 1.057 - 1.138. Assume the value is 1.1, and at the same time assume that the maximum area yu of the allowable risk model is 10 (the unit is determined according to the actual situation);

[0080] Calculate the risk mean value. Assume the calculation result yt1 = 4;

[0081] Extract the area of the risk model at the end time point (t5) of the arrangement as the risk valuation yt2 = S5. Assume yt2 = 3;

[0082] Assume that the influence weight factor of the risk mean value is 0.6, and the influence weight factor of the risk valuation is 0.4; then calculate the potential hazard index σ according to the formula;

[0083] Path optimization: Based on the potential hazard index σ of the current dynamic obstacle to the flight process of the UAV, determine whether it is necessary to re-adjust the preliminary return landing path;

[0084] Specifically:

[0085] S1: Identify the type of the current dynamic obstacle, set the threshold index range of the potential hazard index σ corresponding to different types of dynamic obstacles, match the potential hazard index σ of the current dynamic obstacle to the flight process of the UAV with the threshold index range. If the potential hazard index σ is lower than the threshold index range, push the risk description "The obstacle is far and stable" to the management personnel;

[0086] S2: If the potential hazard index σ is within the threshold index range, push the risk description "The obstacle is approaching but controllable" to the management personnel, and at the same time use the RRT algorithm to generate three candidate paths and push them to the management personnel. After the management personnel receive and replace the corresponding candidate path, adjust the preliminary return landing path;

[0087] S3: If the potential hazard index σ is higher than the threshold index range, push the risk description "The risk impact of the obstacle is high" to the management personnel, and after performing the candidate path push in S2, activate the preset UAV emergency obstacle avoidance settings; The UAV emergency obstacle avoidance settings include vertical avoidance (climbing / descending 5m) and starting emergency braking (maximum deceleration rate);

[0088] It should be noted that if a new obstacle is detected during the flight of the candidate path, repeat the path optimization until the UAV safely reaches the landing point position range;

[0089] Landing Optimization: When the drone reaches the position range of the landing point, it reduces its flight altitude and enters the landing phase. A preset landing pre-preparation duration is set during the landing phase. During the landing pre-preparation duration, the landing evaluation index sera of each group of available landing areas within the landing point position range is analyzed, and the final landing position of the drone is determined based on the landing evaluation index sera of each group of available landing areas;

[0090] It should be noted that

[0091] Lidar: During the landing pre-preparation duration, continuously scan the landing point position range; obtain three-dimensional point cloud data of the surrounding environment at a high frequency (such as 2000 scans per second), with an accuracy up to centimeter level (such as ±5 cm); construct a detailed terrain model through these data to identify information such as the undulation of the ground, the position and shape of obstacles, etc.; for example, using the point cloud data, it is possible to clearly distinguish raised stones, potholes, and small buildings and other obstacles on the ground;

[0092] Vision Sensor: Equipped with a high-definition camera, take images of the landing area from different angles; use advanced image recognition algorithms, such as convolutional neural networks (CNN) based on deep learning, to identify ground textures, color features, etc.; by analyzing the images, different ground materials such as grassland, cement ground, and sandy ground can be distinguished, and at the same time, dynamic obstacles such as moving people and animals can be detected; for example, a trained CNN model can accurately identify pedestrians on the grassland and track their position changes in real time;

[0093] The landing point position range is divided into regions. Areas without obvious obstacles (such as obstacle height less than 0.5 meters and density less than 0.1 per square meter) and relatively flat ground (slope less than 5°) are initially designated as available landing areas;

[0094] Specifically:

[0095] M1: Extract the coverage area of obstacles within each group of available landing areas, and calculate the proportion of the obstacle coverage area in the total area of the available landing area to obtain the obstacle influence value kz1 of each group of available landing areas; obtain the height variance of the point cloud data within each group of available landing areas as the ground influence value kz2 of each group of available landing areas;

[0096] Use an electromagnetic sensor to detect the electromagnetic interference intensity around each group of available landing areas as the electromagnetic interference value kz3 of each group of available landing areas;

[0097] M2: Preset the allowable obstacle value, allowable ground value, and allowable electromagnetic value corresponding to the obstacle influence value kz1, ground influence value kz2, and electromagnetic interference value kz3 respectively, denoted as ks1, ks2, and ks3;

[0098] It should be noted that the allowable obstacle value, the allowable ground value, and the allowable electromagnetic value can be set according to the conditions of the landing area and the model of the UAV.

[0099] M3: According to the formula Perform weighted calculations on the obstacle influence value kz1, the ground influence value kz2, and the electromagnetic interference value kz3 of each group of available landing areas to obtain the landing evaluation index sera of each group of available landing areas; where c1, c2, and c3 are the influence weight factors of the obstacle influence value kz1, the ground influence value kz2, and the electromagnetic interference value kz3 respectively, and the specific values are set according to the weather conditions, such as strong wind and rainfall.

[0100] Strong wind weather: It may cause the UAV to be affected by a relatively large horizontal wind force. At this time, the flatness of the ground has a relatively smaller impact on landing safety, while the interference of obstacles on the UAV's landing path will increase because the strong wind may make it more difficult for the UAV to avoid obstacles. Therefore, the weight c1 of the obstacle influence value kz1 can be appropriately increased, and the weight c2 of the ground influence value kz2 can be reduced at the same time.

[0101] Rainy weather: Rainwater may affect the accuracy of the electromagnetic sensor, resulting in an increase in the measurement error of the electromagnetic interference value kz3. At the same time, it may also make the ground slippery, increasing the impact of the ground conditions on landing; at this time, the weight c3 of the electromagnetic interference value kz3 can be appropriately reduced, and the weight c2 of the ground influence value kz2 can be increased.

[0102] It should be noted that by comprehensively considering various factors such as the obstacle coverage area, the ground flatness, and the electromagnetic interference intensity in the available landing area, the suitability of each available landing area can be comprehensively evaluated, so as to accurately select the safest and most suitable landing position for the UAV, reduce the risks during landing, and improve the success rate and safety of landing.

[0103] It is allowed to set different allowable values according to the conditions of the landing area and the model of the UAV, so that this method can flexibly adapt to various different environments and UAV types, and has high versatility and scalability.

[0104] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the present invention to only the specific implementation manners. Obviously, many modifications and changes can be made according to the content of this specification. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the present invention, so that those skilled in the relevant technical fields can well understand and utilize the present invention. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. A method for Beidou inspection drone return landing and obstacle avoidance, characterized in that: include: Landing optimization: When the drone reaches the landing point, it lowers its flight altitude and enters the landing phase. During the landing phase, a landing preparation time is preset. During the landing preparation time, the landing evaluation index sera of each group of landing areas within the landing point is analyzed, and the final landing position of the drone is determined. Specifically: M1: Extract the coverage area of ​​obstacles in each group of landing areas, and calculate the proportion of the obstacle coverage area in the total area of ​​the landing area to obtain the obstacle impact value kz1 of each group of landing areas; obtain the height variance of the point cloud data in each group of landing areas as the ground impact value kz2 of each group of landing areas; Using electromagnetic sensors to detect the electromagnetic interference intensity around each group of landing areas as the electromagnetic interference value kz3 of each group of landing areas; M2: the allowed obstacle value, allowed ground value and allowed electromagnetic value corresponding to the preset obstacle impact value kz1, ground impact value kz2 and electromagnetic interference value kz3, respectively, recorded as ks1, ks2 and ks3; M3: Comprehensively analyze the obstacle impact value kz1, ground impact value kz2 and electromagnetic interference value kz3 of each group of landing areas to obtain the landing evaluation index sera of each group of landing areas; M4: Select the landing area with the lowest landing evaluation index sera as the final landing position of the drone.

2. A method for Beidou inspection and patrol UAV return landing and obstacle avoidance according to claim 1, characterized in that: The specific analysis of obtaining the landing evaluation index sera of each group of landing areas is as follows: According to the formula The obstacle impact value kz1, ground impact value kz2 and electromagnetic interference value kz3 of each group of landing areas are weightedly calculated to obtain the landing evaluation index sera of each group of landing areas; wherein c1, c2 and c3 are the impact weight factors of the obstacle impact value kz1, the ground impact value kz2 and the electromagnetic interference value kz3 respectively.

3. A method for Beidou inspection and patrol UAV return landing and obstacle avoidance according to claim 2, characterized in that: Also includes: Path planning: Based on the current location of the drone and the landing point, combined with map data and obstacle information, and using the high-precision positioning function of the Beidou positioning system, the initial return landing path for the drone is planned; Real-time obstacle detection: When the UAV is flying along the initial return landing path, the dynamic obstacles within the set distance range of the UAV are detected in real time. When the corresponding dynamic obstacle appears within the set distance range of the UAV at a certain time point, the time point of appearance is first marked as the starting point. After identifying the type of the dynamic obstacle, the movement track of the dynamic obstacle within the set time window after the starting point is analyzed to determine the potential hidden danger index σ of the current dynamic obstacle to the UAV flight process; Path optimization: Based on the potential risk index σ of the current dynamic obstacles to the UAV flight process, determine whether it is necessary to readjust the initial return landing path.

4. A method for Beidou inspection and patrol UAV return landing and obstacle avoidance according to claim 3, characterized in that: The movement trajectory of the dynamic obstacle within the set time window after the starting point is analyzed, specifically: The three-dimensional coordinates of the dynamic obstacle at each time point within the set time window are extracted, and discrete trajectory points are formed. For the three-dimensional coordinates of the drone's location and the trajectory points of the dynamic obstacle at the same time point, the corresponding coordinate points in the three-dimensional space are drawn. After the drawing is completed, the two sets of coordinate points are connected as the hypotenuse of the right triangle; taking the three-dimensional coordinates of the drone's location as the starting point, a perpendicular line to the horizontal line of the trajectory point is constructed as the right angle of the right triangle, and the hypotenuse and the right angle are connected to construct a right triangle. The right triangle constructed at each time point is used as the risk model between the dynamic obstacle and the drone at each time point within the set time window; Extract the area of ​​the risk model at each time point within the set time window and arrange them in chronological order. After the arrangement is completed, calculate the difference in the area of ​​adjacent time points; if the difference at a certain time point is greater than 0, it is marked as a risk increase value; if the difference at a certain time point is less than 0, take the absolute value and mark it as a risk decrease value; After summing up all risk increase values ​​and risk decrease values, we get the total increase value and the total decrease value. Subtract the total decrease value from the total increase value to get the trend judgment value. If the trend judgment value is positive, it is judged as risk increase. Otherwise, it is judged as risk decrease after taking the absolute value. Preset the risk intervals corresponding to the risk appreciation and risk depreciation, and each risk interval of each risk appreciation group corresponds to an additional coefficient for appreciation, and each risk interval of each risk depreciation group corresponds to an additional coefficient for depreciation; The determination result based on the trend determination value calculated within the set time window is converted into an additional coefficient for value increase or a additional coefficient for value reduction, which is denoted as Gi, where i=1 or 2.

5. A method for Beidou inspection and patrol UAV return landing and obstacle avoidance according to claim 4, characterized in that: Analysis of the movement of dynamic obstacles within a set time window after the starting point, including: After identifying the type of dynamic obstacles, a set of stability additional coefficients and maximum areas of the allowed risk model are preset for different types of dynamic obstacles, respectively, which are denoted as λ and yu; The risk model area at each time point in the set time window is averaged to obtain the risk mean, and the area of ​​the risk model at the end time point in the set time window is extracted as the risk estimate; The risk mean and risk estimate between the dynamic obstacle and the UAV within the set time window are marked as yt1 and yt2 respectively.

6. A method for Beidou inspection and patrol UAV return landing and obstacle avoidance according to claim 5, characterized in that: Determine the potential hidden danger index σ of the current dynamic obstacle to the UAV flight process, specifically: According to the formula The risk mean and risk valuation between dynamic obstacles and UAVs within the set time window are weightedly calculated to obtain the potential hazard index σ of the current dynamic obstacles to the UAV flight process; where η1 and η2 are the impact weight factors of the risk mean and risk valuation, respectively.

7. A method for Beidou inspection and patrol UAV return landing and obstacle avoidance according to claim 6, characterized in that: Based on the potential risk index σ of the current dynamic obstacles to the UAV flight process, it is determined whether it is necessary to readjust the initial return landing path, specifically: Identify the type of the current dynamic obstacle, set the threshold index range of the potential hazard index σ corresponding to different types of dynamic obstacles, match the potential hazard index σ of the current dynamic obstacle to the UAV flight process with the threshold index range, and if the potential hazard index σ is lower than the threshold index range, push the risk description "the obstacle is far away and stable" to the management personnel.

8. A method for Beidou inspection and patrol UAV return landing and obstacle avoidance according to claim 7, characterized in that: Determine whether it is necessary to readjust the initial return landing path, including: If the potential hazard index σ is within the threshold index range, the risk description "obstacle approaching but controllable" is pushed to the manager, and three candidate paths are generated using the RRT algorithm and pushed to the manager. After receiving the corresponding candidate path, the manager adjusts the initial return landing path; If the potential hazard index σ is higher than the threshold index range, the risk description "obstacle risk impact is high" is pushed to the manager, and after the candidate path push of S2 is executed, the preset drone emergency obstacle avoidance setting is started.

Images