Method for adjusting safe distance of unmanned aerial vehicle

Abstract

The invention relates to the field of unmanned aerial vehicles, in particular to an unmanned aerial vehicle safety distance adjustment method, which comprises the steps of acquiring a flight video and positioning data of an unmanned aerial vehicle, constructing an obstacle set and training a target detection model, acquiring position coordinates of historical obstacles, predicting the positions of the obstacles by using an LSTM model, and calculating prediction accuracy. Inputting the real-time frame image into the target detection model to obtain the position coordinate of the obstacle, predicting the position information of the next moment, and calculating the direction included angle and the trajectory overlapping probability of the unmanned aerial vehicle and the obstacle; calculating a relative speed according to the speed vectors and the direction included angles of the unmanned aerial vehicle and the obstacle at the current moment; and dynamically adjusting the safety distance of the unmanned aerial vehicle according to the relative speed between the unmanned aerial vehicle and the obstacle, the trajectory overlapping probability and the prediction accuracy. By dynamically adjusting the safe distance, the unmanned aerial vehicle can maintain safe flight, and meanwhile, the operation reliability and the practicability of the unmanned aerial vehicle are improved.

Description

Technical Field

[0001] The present invention relates to the field of unmanned aerial vehicles, and more specifically, to a method for adjusting the safety distance of an unmanned aerial vehicle. Background Art

[0002] As drones become more popular, regulations and regulatory measures have been set for drone flights to ensure the safety and legal use of drones. These regulations usually involve aspects such as the flight altitude and flight area of ​​drones, the qualification requirements of operators, and privacy protection. At the same time, the safe operation and privacy protection of drones are the focus of public attention. Drones may infringe on personal privacy, cause mid-air collisions with manned aircraft, or be used for illegal activities. Therefore, the operation of drones needs to strictly comply with relevant regulations and take appropriate safety measures.

[0003] The existing Chinese patent application document with publication number CN117826843A discloses a method and system for intelligent obstacle avoidance of unmanned aerial vehicles based on three-dimensional point clouds, the method comprising: obtaining a three-dimensional point cloud data set of a power transmission engineering scene and constructing a moving sliding window, calculating the outlier value and the wire edge performance value of the target point in the three-dimensional point cloud data set in the moving sliding window, and judging to obtain a first signal-to-noise value; calculating the outlier density value of the target point to obtain a second signal-to-noise value, adaptively adjusting the moving sliding window according to the second signal-to-noise value and performing denoising to obtain a significant data set, and using the significant data set to build a power transmission engineering site model to guide the unmanned aerial vehicle to perform intelligent obstacle avoidance. The present invention improves the adaptability of denoising of three-dimensional point cloud data, thereby more accurately denoising to obtain a more accurate model of the power transmission engineering site, thereby improving the accuracy and efficiency of the unmanned aerial vehicle intelligent obstacle avoidance method.

[0004] This application document only processes drone data and provides accurate analysis data for drone obstacle avoidance. Currently, drones are equipped with laser rangefinders to assist in positioning and obstacle avoidance, but drone obstacle avoidance still relies on manual control. At the same time, when manually operating drones to avoid obstacles, they may encounter problems such as insufficient reaction time, visual blind spots, and complex environments, which may lead to obstacle avoidance failures and increase the risk of obstacle avoidance caused by operational errors. Summary of the invention

[0005] In order to solve the problem that when manually operating a drone to avoid obstacles, insufficient reaction time, visual blind spots and complex environments may be encountered, which may lead to obstacle avoidance failure, the present invention provides solutions in the following aspects.

[0006] A method for adjusting the safety distance of an unmanned aerial vehicle comprises: obtaining a video and positioning data of a flight mission route of the unmanned aerial vehicle, collecting and annotating frame images of the video, constructing an obstacle set, obtaining a position coordinate sequence of the unmanned aerial vehicle according to the positioning data in time sequence, training a target detection model using the obstacle set, obtaining the position coordinates of historical obstacles, predicting the position information of the next moment based on the position coordinates using an LSTM model, and calculating the prediction accuracy; inputting a real-time frame image into the target detection model to obtain the position coordinates of the obstacle, predicting the position information of the next moment, calculating the direction angle between the unmanned aerial vehicle and the obstacle at the previous moment and the current moment, and the probability of trajectory overlap within a preset time before the current moment and the next moment meet; calculating the relative speed according to the speed vector and the direction angle of the unmanned aerial vehicle and the obstacle at the current moment respectively; dynamically adjusting the safety distance of the unmanned aerial vehicle according to the relative speed between the unmanned aerial vehicle and the obstacle, the probability of trajectory overlap and the prediction accuracy; wherein the safety distance Satisfies the following relationship: , where Indicates the initial safety distance. Indicates The relative speed between the drone and the obstacle at that moment, represents the overlap probability between the trajectory of the drone and the obstacle, Indicates the prediction accuracy of the LSTM model for the obstacle position coordinates, Indicates The predicted distance between the drone and obstacles at all times.

[0007] The effect is: by real-time analysis of video streams and positioning data, the drone can accurately identify and predict the location of obstacles, thereby effectively avoiding obstacles and reducing collision risks. It can dynamically adjust according to the relative speed between the drone and the obstacle, the probability of trajectory overlap and the prediction accuracy, so that the safe distance is more adapted to the current flight environment and conditions, improving the flexibility and safety of the flight. By calculating the directional angle and relative speed between the drone and the obstacle, the drone can analyze the relative motion state of the two in real time, predict potential collision points and times, and avoid obstacles in advance.

[0008] Preferably, the target detection model is a Yolov5 network model, the input is an obstacle set, and the output is the position and confidence of the obstacle.

[0009] Preferably, the input of the LSTM model is a sequence of position coordinates of the obstacle, and the output is the position coordinates of the obstacle at the next moment, wherein the position coordinates are the centroid point of the rectangular candidate box at the outermost edge of the obstacle.

[0010] The effect is that by using the centroid point of the rectangular candidate box at the outermost edge of the obstacle as the position coordinate, the position of the obstacle can be more accurately represented. By accurately predicting the position of the obstacle, the drone can make obstacle avoidance decisions in advance, thereby reducing the risk of collision and improving flight safety.

[0011] Preferably, the prediction accuracy includes: Obtain the predicted position coordinates and confidence of the obstacle at each historical moment, calculate the distance difference between the predicted position coordinates and the actual position coordinates of the obstacle, take the cumulative sum of the confidence and the corresponding distance difference as the cumulative error of all historical moments, and use the ratio of the cumulative error to the total time of the flight mission to calculate the distance difference between the predicted position coordinates and the actual position coordinates of the obstacle. The prediction accuracy is calculated using a negative exponential function with as the base.

[0012] Preferably, the prediction accuracy further includes: Calculate the sum of squares of the differences between all true position coordinates and the mean of all true position coordinates during the historical flight time to obtain the total square error; calculate the sum of squares of the differences between all true position coordinates and the predicted position coordinates to obtain the residual square sum; take the complement of the ratio between the residual square sum and the total square error as the determination coefficient; The product of the average value of the built-in confidence of all historical flight times and the coefficient of determination is calculated and normalized to obtain the prediction accuracy.

[0013] Preferably, the direction angle includes: Establish a three-dimensional coordinate system, and determine the position coordinates of the drone in the three-dimensional coordinate system and the position coordinate sequence of the obstacles according to the drone positioning data, where the initial position of the drone is taken as the coordinate origin and the initial flight direction of the drone is axis; Calculate the velocity vector and the modulus of the velocity vector of the UAV and the obstacle from the previous moment to the current moment respectively, and obtain the unit velocity vector of the UAV and the unit velocity vector of the obstacle; Calculate the ratio between the product of the unit velocity vector of the drone and the unit velocity vector of the obstacle and the product of the modulus of the corresponding velocity vectors, and use the arc cosine function to obtain the direction angle between the drone and the obstacle at the current moment.

[0014] The effect is: by establishing a three-dimensional coordinate system and calculating the unit velocity vectors of the drone and the obstacle, the directional angle between the two can be accurately determined. By determining the directional angle between the drone and the obstacle, the drone can more accurately judge how to adjust its flight path to avoid collision. Real-time calculation of the directional angle enables the drone to respond quickly to environmental changes, adjust its flight direction in time, and avoid potential collisions.

[0015] Preferably, the trajectory overlap probability includes: Input the current obstacle position coordinates into the LSTM model, and make rolling predictions to obtain the obstacle position coordinate sequences at multiple preset moments. The rolling prediction is: The coordinate information of the obstacle at the moment is input into the LSTM model to obtain the The predicted coordinate information at the moment The predicted coordinate information at each moment is input into the LSTM model and iterated to obtain the position coordinate sequence of the obstacle at multiple preset moments. The Euclidean distance between the position coordinate sequence of the obstacle at multiple preset moments and the position coordinate of the corresponding drone is calculated. If the Euclidean distance is greater than the preset distance threshold, there is no overlap. Otherwise, the ratio of the number of overlapping coordinate points to the total number of trajectory points is counted to obtain the overlap probability.

[0016] The effect is: through rolling prediction, the drone can predict the position of obstacles at multiple moments in the future, so as to make obstacle avoidance decisions in advance. The rolling prediction provides a dynamic trajectory of the obstacle position changing over time, allowing the drone to analyze and predict potential collision risks. By counting the number of overlapping coordinate points, the drone can calculate the probability of overlapping with the obstacle trajectory, thereby enhancing flight safety.

[0017] Preferably, the relative speed satisfies the following relationship: ; In the formula, express The relative speed between the drone and the obstacle at that moment, express The speed of the drone at the moment, express The speed of the obstacle at any moment, express The angle between the drone and the obstacle at that moment, Represents the cosine function.

[0018] The effect is that by calculating the relative speed, the relative motion between the drone and the obstacle can be accurately analyzed, which is helpful for predicting collision risks and planning obstacle avoidance strategies.

[0019] The present invention has the following effects: 1. The present invention reduces the reliance on manual operation through an automated safety distance adjustment method, thereby reducing the risk of collision caused by human negligence or error. By dynamically adjusting the safety distance, the drone can perform tasks more effectively while maintaining safe flight, thereby improving the reliability of operation and the practicality of the drone.

[0020] 2. By using the Yolov5 network model and the LSTM model, the drone can detect obstacles in real time and predict their future positions. This automated prediction and obstacle avoidance mechanism significantly reduces the reaction time of manual operation, improves the speed and accuracy of obstacle avoidance decisions, and effectively avoids collisions caused by insufficient reaction time in complex environments.

[0021] 3. The present invention can more accurately assess the potential collision risk by calculating the direction angle and relative speed between the drone and the obstacle. Combined with the calculation of trajectory overlap probability, the drone can identify high-risk areas in advance and adjust the flight path in time, enhancing flight safety in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein: Figure 1 It is a method flow chart of steps S1 to S3 in a method for adjusting the safety distance of a drone according to an embodiment of the present invention. DETAILED DESCRIPTION

[0023] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0024] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0025] Reference Figure 1 A method for adjusting the safety distance of a drone includes steps S1 to S3, which are as follows: S1: Obtain the video and positioning data of the UAV flight mission route, collect frame images of the video and annotate them, build an obstacle set, obtain the position coordinate sequence of the UAV in time sequence for the positioning data, use the obstacle set to train the target detection model, obtain the position coordinates of historical obstacles, use the LSTM model to predict the position information at the next moment based on the position coordinates, and calculate the prediction accuracy.

[0026] To further illustrate, the flight mission route videos under different environments are obtained, and frames are extracted from the videos to obtain a frame image every 30 frames. Obstacles in the frame images are marked by box selection to construct an obstacle set.

[0027] The target detection model is the Yolov5 network model, the input is the obstacle set, and the output is the location and confidence of the obstacle.

[0028] The input of the LSTM model is the position coordinate sequence of the obstacle, and the output is the position coordinate of the obstacle at the next moment, where the position coordinate is the centroid point of the rectangular candidate box at the outermost edge of the obstacle.

[0029] It should be noted that the Yolov5 network model and the LSTM model are existing well-known technologies and will not be described in detail. The centroid point of the rectangular candidate box is used as the position coordinate of the obstacle to simplify the calculation of trajectory prediction and collision test, especially when the obstacle has an irregular shape. At the same time, it can reduce the error caused by the change of the edge position of the rectangular box, making the prediction of the LSTM model more stable and reliable.

[0030] Obtain the predicted position coordinates and confidence of the obstacle at each historical moment, calculate the distance difference between the predicted position coordinates and the actual position coordinates of the obstacle, take the sum of the confidence and the corresponding distance difference as the cumulative error of all historical moments, and use the ratio of the cumulative error to the total time of the flight mission to calculate the distance difference between the predicted position coordinates and the actual position coordinates. The prediction accuracy is calculated using a negative exponential function with as the base.

[0031] Specifically, the prediction accuracy satisfies the following relationship: ; In the formula, Indicates the prediction accuracy of the LSTM model for the obstacle position coordinates, Indicates the total time the drone performs the flight mission. Indicates The confidence of the obstacle candidate box at the moment, Indicates obstacle The distance difference between the predicted position coordinates and the actual position coordinates at the moment, Indicates An exponential function with base .

[0032] That is to say, the prediction accuracy range is When the prediction accuracy approaches 1, it indicates that the prediction accuracy is higher, otherwise, it indicates that the prediction accuracy is lower.

[0033] In addition, in another embodiment, the sum of squares of differences between all real position coordinates and the mean of all real position coordinates within the historical flight time is calculated to obtain a total square error; the sum of squares of differences between all real position coordinates and predicted position coordinates is calculated to obtain a residual square sum; the complement of the ratio between the residual square sum and the total square error is used as a determination coefficient; The product of the average value of the built-in confidence of all historical flight times and the coefficient of determination is calculated and normalized to obtain the prediction accuracy.

[0034] Specifically, the coefficient of determination satisfies the following relationship: ; In the formula, represents the determination coefficient of the LSTM model, represents the total time of historical flight, Indicates The real position coordinates of the obstacle at the moment, Indicates The predicted position coordinates of the obstacle at time, Indicates historical flights The mean of the real position coordinates of the obstacle within time.

[0035] Specifically, the prediction accuracy satisfies the following relationship: ; In the formula, Indicates the prediction accuracy of the LSTM model for the obstacle position coordinates, is the normalization function, represents the determination coefficient of the LSTM model, represents the total time of historical flight, Indicates The confidence level of the obstacle at the moment.

[0036] S2: Input the real-time frame image into the target detection model to obtain the position coordinates of the obstacle, and predict the position information at the next moment, calculate the direction angle between the drone and the obstacle at the previous moment and the current moment, and the probability of trajectory overlap within the preset time before the current moment meets the next moment.

[0037] Establish a three-dimensional coordinate system, and determine the position coordinates of the drone in the three-dimensional coordinate system and the position coordinate sequence of the obstacles according to the drone positioning data, where the initial position of the drone is taken as the coordinate origin and the initial flight direction of the drone is axis; Calculate the velocity vector and the modulus of the velocity vector of the UAV and the obstacle from the previous moment to the current moment respectively, and obtain the unit velocity vector of the UAV and the unit velocity vector of the obstacle; Calculate the ratio between the product of the unit velocity vector of the drone and the unit velocity vector of the obstacle and the product of the modulus of the corresponding velocity vectors, and use the arc cosine function to obtain the direction angle between the drone and the obstacle at the current moment.

[0038] To further illustrate, by setting the initial position of the UAV as the coordinate origin and the initial flight direction of the UAV as the X-axis, a reference frame is provided for the position of the UAV and the obstacle. By analyzing the directional angle between the UAV and the obstacle, it is helpful to operate the UAV for obstacle avoidance and path planning, effectively reducing the risk of collision.

[0039] Specifically, the direction angle satisfies the following relationship: ; In the formula, Indicates The angle between the drone and the obstacle at that moment, , Respectively represent The unit velocity vector of the drone and the obstacle at this moment, , Respectively represent The modulus of the unit vector of the velocity of the drone and the obstacle at the moment, Represents the conversion formula for converting the direction angle from radians to degrees, is the inverse cosine function.

[0040] It should be noted that Indicates the similarity between two unit vectors in terms of quantity, and its value range is , where the value is When , it means that the vectors are exactly the same, and the value is When , it means that the vectors are completely opposite, and the value is When , it means the vectors are perpendicular; Used to convert angles from radians to degrees. The relationship between radians and degrees is: , to convert the angle from radians to degrees, .

[0041] Input the current obstacle position coordinates into the LSTM model, and make rolling predictions to obtain the obstacle position coordinate sequences at multiple preset moments. The rolling prediction is: The coordinate information of the obstacle at the moment is input into the LSTM model to obtain the The predicted coordinate information at the moment The predicted coordinate information at each moment is input into the LSTM model and iterated to obtain the position coordinate sequence of the obstacle at multiple preset moments. The Euclidean distance between the position coordinate sequence of the obstacle at multiple preset moments and the position coordinate of the corresponding drone is calculated. If the Euclidean distance is greater than the preset distance threshold, there is no overlap. Otherwise, the ratio of the number of overlapping coordinate points to the total number of trajectory points is counted to obtain the overlap probability.

[0042] It should be noted that rolling prediction is a time series prediction method that uses the recursive nature of the LSTM model to predict the position information of multiple time steps in the future through continuous iteration. This method can capture long-term dependencies in time series data. By inputting the location coordinates of the obstacle at the current moment into the LSTM model, the obstacle position prediction for a series of time points in the future can be obtained. This helps predict the potential path of the obstacle.

[0043] Specifically, the overlap probability satisfies the following relationship: ; In the formula, represents the overlap probability between the trajectory of the drone and the obstacle, Indicates the number of overlapping points between the drone’s and obstacles’ trajectories, Indicates the total number of trajectory points between the drone and obstacles.

[0044] Exemplarily, the preset distance threshold is , which can be adjusted according to the actual situation. The preset time is 10 moments before the current time, including the current time. .

[0045] S3: Calculate the relative speed based on the velocity vectors and direction angles of the drone and obstacle at the current moment; dynamically adjust the drone's safety distance based on the relative speed between the drone and the obstacle, the trajectory overlap probability, and the prediction accuracy.

[0046] Specifically, the relative speed satisfies the following relationship: ; In the formula, express The relative speed between the drone and the obstacle at that moment, express The speed of the drone at the moment, express The speed of the obstacle at any moment, express The angle between the drone and the obstacle at that moment, Represents the cosine function.

[0047] Specifically, the safety distance satisfies the following relationship: ; In the formula, express The safe distance of drones at all times, Indicates the initial safety distance. Indicates The relative speed between the drone and the obstacle at that moment, represents the overlap probability between the trajectory of the drone and the obstacle, Indicates the prediction accuracy of the LSTM model for the obstacle position coordinates, Indicates The predicted distance between the drone and obstacles at all times.

[0048] By considering relative speed, trajectory overlap probability and prediction accuracy, the UAV can dynamically adjust the safety distance to adapt to the changing flight environment. The dynamically adjusted safety distance helps the UAV make more accurate obstacle avoidance decisions when approaching obstacles, thereby improving flight safety.

[0049] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0050] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A method for adjusting the safety distance of an unmanned aerial vehicle, characterized in that: include: Obtain the video and positioning data of the UAV flight mission route, collect and annotate frame images of the video, build an obstacle set, obtain the position coordinate sequence of the UAV according to the time sequence of the positioning data, use the obstacle set to train the target detection model, obtain the position coordinates of historical obstacles, use the LSTM model to predict the position information at the next moment based on the position coordinates, and calculate the prediction accuracy; Input the real-time frame image into the target detection model to obtain the position coordinates of the obstacle, and predict the position information at the next moment, calculate the direction angle between the drone and the obstacle at the previous moment and the current moment, and the probability of trajectory overlap within the preset time before the current moment and the next moment meet; The relative speed is calculated based on the velocity vector and direction angle of the drone and obstacle at the current moment; the safety distance of the drone is dynamically adjusted based on the relative speed, trajectory overlap probability and prediction accuracy between the drone and the obstacle; Among them, the safety distance Satisfies the following relationship: , where Indicates the initial safety distance. Indicates The relative speed between the drone and the obstacle at that moment, represents the overlap probability between the trajectory of the drone and the obstacle, Indicates the prediction accuracy of the LSTM model for the obstacle position coordinates, Indicates The predicted distance between the drone and obstacles at all times.

2. The method for adjusting the safety distance of an unmanned aerial vehicle according to claim 1, characterized in that: The target detection model is a Yolov5 network model, the input is an obstacle set, and the output is the location and confidence of the obstacle.

3. The method for adjusting the safety distance of an unmanned aerial vehicle according to claim 1, characterized in that: The input of the LSTM model is a sequence of position coordinates of the obstacle, and the output is the position coordinates of the obstacle at the next moment, wherein the position coordinates are the centroid point of the rectangular candidate box at the outermost edge of the obstacle.

4. The method for adjusting the safety distance of an unmanned aerial vehicle according to claim 1, characterized in that: The prediction accuracy includes: Obtain the predicted position coordinates and confidence of the obstacle at each historical moment, calculate the distance difference between the predicted position coordinates and the actual position coordinates of the obstacle, take the cumulative sum of the confidence and the corresponding distance difference as the cumulative error of all historical moments, and use the ratio of the cumulative error to the total time of the flight mission to calculate the distance difference between the predicted position coordinates and the actual position coordinates of the obstacle. The prediction accuracy is calculated using a negative exponential function with as the base.

5. The method for adjusting the safety distance of an unmanned aerial vehicle according to claim 1, characterized in that: The prediction accuracy also includes: Calculate the sum of squares of the differences between all true position coordinates and the mean of all true position coordinates during the historical flight time to obtain the total square error; calculate the sum of squares of the differences between all true position coordinates and the predicted position coordinates to obtain the residual square sum; take the complement of the ratio between the residual square sum and the total square error as the determination coefficient; The product of the average value of the built-in confidence of all historical flight times and the coefficient of determination is calculated and normalized to obtain the prediction accuracy.

6. The method for adjusting the safety distance of an unmanned aerial vehicle according to claim 1, characterized in that: The direction angle includes: Establish a three-dimensional coordinate system, and determine the position coordinates of the drone in the three-dimensional coordinate system and the position coordinate sequence of the obstacles according to the drone positioning data, where the initial position of the drone is taken as the coordinate origin and the initial flight direction of the drone is axis; Calculate the velocity vector and the modulus of the velocity vector of the UAV and the obstacle from the previous moment to the current moment respectively, and obtain the unit velocity vector of the UAV and the unit velocity vector of the obstacle; Calculate the ratio between the product of the unit velocity vector of the drone and the unit velocity vector of the obstacle and the product of the modulus of the corresponding velocity vectors, and use the arc cosine function to obtain the direction angle between the drone and the obstacle at the current moment.

7. The method for adjusting the safety distance of an unmanned aerial vehicle according to claim 1, characterized in that: The trajectory overlap probability includes: Input the current obstacle position coordinates into the LSTM model, and make rolling predictions to obtain the obstacle position coordinate sequences at multiple preset moments. The rolling prediction is: The coordinate information of the obstacle at the moment is input into the LSTM model to obtain the The predicted coordinate information at the moment The predicted coordinate information at each moment is input into the LSTM model and iterated to obtain the position coordinate sequence of the obstacle at multiple preset moments. The Euclidean distance between the position coordinate sequence of the obstacle at multiple preset moments and the position coordinate of the corresponding drone is calculated. If the Euclidean distance is greater than the preset distance threshold, there is no overlap. Otherwise, the ratio of the number of overlapping coordinate points to the total number of trajectory points is counted to obtain the overlap probability.

8. The method for adjusting the safety distance of an unmanned aerial vehicle according to claim 1, characterized in that: The relative speed satisfies the following relationship: ； In the formula, express The relative speed between the drone and the obstacle at that moment, express The speed of the drone at the moment, express The speed of the obstacle at any moment, express The angle between the drone and the obstacle at that moment, Represents the cosine function.

Images