Urban low-altitude unmanned aerial vehicle tactical conflict detection solution path optimization method

Abstract

The invention belongs to the field of path planning, and provides an urban low-altitude unmanned aerial vehicle tactical conflict detection solution path optimization method. Relates to the technical field of unmanned aerial vehicle flight safety. The method comprises the following steps: firstly, receiving an ADS-B message through the unmanned aerial vehicle to analyze longitude and latitude, height and speed information, and obtaining the position of the unmanned aerial vehicle relative to a fixed point on the ground through coordinate conversion; generating a collision cone conflict detection model area according to the flight position and speed of the unmanned aerial vehicle and the protection area; secondly, establishing a reciprocal speed obstacle region based on a collision cone model, obtaining a new reciprocal speed obstacle region by considering the collision risk degree, and realizing conflict solution through the speed outside the region; and meanwhile, the conflict solution speed is further optimized by utilizing a gradient descent principle. According to the method, a protection area radius adjustment function and a gradient descent algorithm thought are fused in an original reciprocal speed obstacle method, unmanned aerial vehicle conflict solution is successfully achieved, and the flight path length of the unmanned aerial vehicle is shortened as much as possible.

Description

Technical Field

[0001] The present invention relates to the field of unmanned aerial vehicle flight safety technology, and in particular to a path optimization method for detecting and resolving tactical conflicts of low-altitude unmanned aerial vehicles in urban areas. Background Art

[0002] This section merely provides background information related to the present disclosure and is not necessarily prior art.

[0003] In recent years, numerous major cities across China have been granted low-altitude flight pilot programs, marking a significant step forward in the development of my country's urban transportation system, providing policy support for the development of Urban Air Mobility (UAM). As urban airspace continues to open up, unmanned aerial vehicles (UAVs) are widely used in urban logistics, aerial photography, traffic monitoring, and other tasks, and their number is growing exponentially. This increasing number of drones inevitably leads to a decrease in available airspace within urban areas, making flight conflicts inevitable. This increases the likelihood of mid-air collisions, which can lead to drone crashes and endanger the safety of pedestrians on the ground.

[0004] To address flight conflicts between multiple low-altitude aircraft operating in the same limited airspace, numerous researchers have begun researching conflict detection and resolution (CD&R) technology to ensure that aircraft can autonomously resolve conflicts when faced with a conflict threat. Typically, each pair of drones determines whether a conflict has occurred and, if so, implements conflict resolution measures. UAV conflict detection and resolution methods can be broadly categorized into two types: strategic-level global trajectory planning and tactical-level localized conflict detection and resolution. CD&R generally refers to the latter, requiring drones to efficiently detect and resolve conflicts in real time within complex and changing environments. When drones are operating autonomously and uncontrolled, they must be assigned a safe flight strategy to prevent collisions with various fixed and moving obstacles. Furthermore, when numerous drones operate within the same limited urban airspace, avoiding moving obstacles can lead to poor flight stability, resulting in fluctuations or oscillations in flight direction, known as "jitter." The risk of conflict or even collision must be considered in advance to determine whether urgent conflict resolution measures are necessary. Alternatively, traditional conflict resolution methods can be optimized and improved to achieve smoother and safer flight paths. At the same time, the length of the drone's flight path should be shortened as much as possible on the basis of conflict resolution to reduce the waste of airspace resources.

[0005] Therefore, in order to ensure that UAVs can fly safely and stably in complex urban environments, it is particularly important to develop an efficient flight conflict detection and resolution algorithm. Currently, the methods for conflict detection and resolution for UAV flights include but are not limited to the dynamic window method, artificial potential field method, speed barrier method and geometric optimization method. In particular, the speed barrier method has been widely used by scholars because its algorithm model is simple and easy to implement compared to other algorithm models. However, when using the traditional speed barrier method and its variant (reciprocal speed barrier method) during the flight of unmanned aerial vehicles, there will be frequent flight jitter problems, and it cannot guarantee that the flight path for flight conflict resolution is short. It does not consider issues such as path length optimization, which deserves further optimization by scholars. Summary of the Invention

[0006] Purpose of the Invention: To address the shortcomings of existing technologies, this paper provides a path optimization method for detecting and resolving tactical conflicts between UAVs in urban low-altitude airspace, based on the CCR-GDO-RVO (Consider Collision Risk-Gradient Descent Optimization-Reciprocal Velocity Obstacle) approach. This method addresses the issues of the velocity-obstacle approach, which often results in severe jitter during flight and fails to consider path length optimization. This method enables the detection and resolution of UAV flight conflicts in low-altitude urban airspace.

[0007] Technical Solution: A method for optimizing the path of tactical conflict detection and resolution for low-altitude unmanned aerial vehicles in urban areas, comprising the following steps:

[0008] Step 1: Each drone is equipped with a small Automatic Dependent Surveillance-Broadcast (ADS-B) receiver that continuously receives ADS-B messages and parses the ADS-B longitude and latitude coordinates through Compact Position Reporting (CPR). It then performs real-time coordinate conversion to obtain the drone's real-time relative ground position.

[0009] Step 1-1: Obtain ADS-B data message, which includes the drone's position and speed.

[0010] Step 1-2: Decode the ADS-B data. Decode the longitude, latitude, altitude, and speed of the UAV in the binary ADS-B data message obtained in step 1-1. Convert the corresponding bits of the message to decimal to obtain the altitude and speed information. Decode the longitude and latitude information by decoding the CPR of the message. Convert the ADS-B data message to a readable text format.

[0011] Step 1-3: Perform coordinate conversion on the decoded ADS-B position data obtained in step 1-2, converting the WGS-84 ellipsoid coordinates of the drone's ADS-B position and the reference position into Earth-centered Earth-fixed coordinates (X, Y, Z). Calculate the East-North-Up (ENU) coordinates (x, y, z) of the drone relative to the ground fixed reference point using the Earth-centered Earth-fixed coordinates (X, Y, Z).

[0012] Step 2: Establish a flight protection zone for drones based on their flight speed and airspace building layout, and build a collision cone model for flight conflict detection.

[0013] Step 2-1: Establish a flight protection zone model for the UAV and determine the flight information.

[0014] Define R 3 It is a three-dimensional real vector space, and the set of drone spherical protection zones V is as follows:

[0015]

[0016] Among them, d h Indicates the radius of the drone's spherical flight protection zone.

[0017] Assume that drone A and drone B are in a limited low-altitude urban airspace. Through ADS-B message analysis, the flight speeds of drone A and drone B are obtained as u1 and u2 respectively. With drone A as the origin O as the ground fixed reference point, line AB as the x-axis, and the direction of vector AB as the positive direction of the x-axis, a three-dimensional space coordinate system is established using the right-hand rule. In the three-dimensional spherical coordinate system, the flight speeds v1 and v2 of the two drones are obtained as shown in the following formula:

[0018] v i =(u i cosφ i sinθ i ,u i sinφ i sinθ i ,u i cosθ i ),i∈1,2;

[0019] Among them, φ i Represents the positive angle between the UAV's flight direction and the x-axis; θ i Represents the positive angle between the drone's flight direction and the z-axis.

[0020] The corresponding protection zone radii of UAV A and UAV B are and

[0021] Assume that the displacement generated by UAV A or UAV B during the kth time period Δt is Then the displacement of the UAV during the k+1th time period Δt is as follows:

[0022]

[0023] Define the maximum speed constraint v max and the maximum acceleration constraint a max Limit the available speed and acceleration of the drone as shown in the following formula:

[0024] ||v i ||≤v max ,i∈1,2;

[0025] ||a i ||≤a max ,i∈1,2;

[0026] Among them, ||.|| represents the vector modulus or Euclidean distance, v i is the speed of the i-th UAV, a i is the acceleration of the i-th UAV.

[0027] Step 2-2: Use the collision cone model to obtain the flight conflict detection area.

[0028] In three-dimensional space, the flight speed of UAV A is v1, the flight speed of UAV B is v2, α is the angle between the line AB between the two UAVs and the boundary of the speed obstacle zone, that is, half of the collision cone vertex angle, γ is the angle between the line AB between the two UAVs and the relative speed v R The angle between them.

[0029] Define the relative collision cone (RCC) model, that is, the relative velocity v at which drones may collide R = the velocity set of v1-v2, as shown in the following formula:

[0030]

[0031] in, is the relative velocity v R The ray drawn at the particle point of drone A; Sphere_B is the flight safety protection zone of drone B, and the collision cone boundary is tangent to Sphere_B.

[0032] Step 2-3: Determine whether a flight conflict occurs through the collision cone model.

[0033] The collision cone model only considers the positional relationship and current state between drones. Taking drone B as the reference point, drone A moves relative to drone B. If the movement speed of drone A relative to drone B is within the collision cone, a flight conflict will occur between the two drones; otherwise, there will be no flight conflict.

[0034] Based on the collision cone model, the following geometric judgment method is used to detect the flight conflict between two drones: when α>γ, there is a flight conflict between the two drones, otherwise there is no flight conflict; the expressions of α and γ are shown in the following formula:

[0035]

[0036] Among them D AB is the relative distance between UAV A and UAV B, D AB is vector AB.

[0037] Step 3: Use the Reciprocal Velocity Obstacle (RVO) conflict detection algorithm to determine the conflict area, assign the UAV a flight speed, and resolve the UAV flight conflict.

[0038] Step 3-1: Obtain the reciprocal speed barrier area through the collision cone area and the flight speed of the drone.

[0039] Taking two drones as an example, the traditional speed obstacle method is iterated bidirectionally to obtain the reciprocal speed obstacle method. The reciprocal speed obstacle conflict detection algorithm considers the reciprocal motion of drone A and drone B at the same time, so that the two drones share the responsibility of obstacle avoidance. The RVO area expression generated by drone A relative to drone B can be obtained:

[0040]

[0041] in, Represents Minkowski and.

[0042] The next moment's flight speed v of UAV A in the area satisfies the following formula:

[0043]

[0044] Where t is a certain flight time in the future, and Sphere_B is the flight safety protection zone of UAV B.

[0045] If there are n drones flying in the airspace, the flight conflict resolution speed set for drone A can be expressed as follows:

[0046]

[0047] Step 3-2: Assign the drone a flyable speed to resolve flight conflicts of the drone.

[0048] Set an optimal flight speed for drone A As shown in the following formula:

[0049]

[0050] Where, represents the magnitude of the optimal flight rate of drone A, which is valued according to the model and its own attributes of the drone. Different types of drones have different optimal flight rates. P1 represents the current position of drone A, represents the end position of drone A.

[0051] Drone A selects a speed vector closest to the optimal flight speed outside the RVO area as the new flight speed As shown in the following formula:

[0052]

[0053] Where, v is the set of flight speeds of drone A at the next moment.

[0054] Step 4: Based on the RVO algorithm, further reduce the conflict occurrence area through the ConsiderCollision Risk-Reciprocal Velocity Obstacle (CCR-RVO) conflict detection algorithm that takes into account the collision risk. Set a flight safety protection area radius adjustment function to adjust the size of the protection area radius, and adjust the flight speed of the drone according to the adjusted protection area radius to adjust the flight conflict resolution speed of the drone.

[0055] Step 4-1: Consider the collision risk between two drones based on their relative positions and speeds, and select an appropriate protection area radius for the drone in real time.

[0056] Increase the range of optional obstacle avoidance speed vectors for the drone at each moment according to the real-time change of the protection area radius. The degree of collision risk between drones can be evaluated based on the gradient-colored collision cone area. As the distance between drone A and drone B gradually decreases, the color of the collision cone area gradually changes from white to gray and then to black. At this time, the degree of collision risk for the two drones gradually increases, and the possibility of collision also increases.

[0057] Set the change ratio coefficient k of the protection area radius, and 0 < k ≤ 1. By reducing the protection area radius in the traditional collision cone at the next flight moment, and then reducing the RCC area, the generated RVO area at this time is the reciprocal velocity obstacle area considering the collision risk; thereby increasing the speed area available for the drone to choose.

[0058] Similarly, if the relative distance is close, a collision will occur soon. In this case, the radius of the protection zone will be expanded at the next flight moment, thereby expanding the RCC and reducing the speed range available for the drone. When the relative distance decreases and approaches the protection zone radius, the protection zone radius remains Remain unchanged.

[0059] Step 4-2: Adjust the UAV's flight conflict resolution speed.

[0060] Drone A expands the range of optional flight conflict resolution speeds, thereby selecting a more optimal flight conflict resolution speed and reducing frequent jitter caused by the large speed barrier area in traditional algorithms, which results in a smaller set of optional speeds and a large number of drones in the airspace.

[0061] Set an optimal flight speed for drone A As shown in the following formula:

[0062]

[0063] UAV A selects a velocity vector closest to the optimal flight speed outside the CCR-RVO area as the new flight speed of UAV A, as shown in the following formula:

[0064]

[0065] Similarly, the flight conflict resolution speed of UAV B is obtained.

[0066] When multiple drones are flying, the flight conflict resolution speed of drone A in step 4-2 is calculated using the following formula:

[0067]

[0068] Where n is the number of drones flying in the airspace.

[0069] Furthermore, the specific method for setting the protection zone radius in step 4 is:

[0070] A regulation function is selected for the proportional coefficient of the protected area radius change. The independent variable of this function is the distance between the two drones, and the dependent variable is the proportional coefficient. The function stipulates that the smaller the distance between the drone and the other drone or obstacle, the larger the k value. As the collision risk increases, the growth rate of the k value gradually increases, that is, the slope of the regulation function at k gradually increases. Conversely, the larger the distance between the two drones, the smaller the k value, and the rate of decrease of the k value gradually decreases, that is, the slope of the regulation function at k gradually decreases.

[0071] The Sigmoid function has the function value increasing as the independent variable increases, and when the independent variable increases, the slope at the corresponding function point also gradually increases, and the slope change of the point on the function curve is continuous. The Sigmoid function expression is shown as follows:

[0072]

[0073] Set the proportionality coefficient k for the change in the radius of the protection area, where 0 < k ≤ 1. Based on the requirements of the proportionality coefficient k and the characteristics of the Sigmoid function, select the adjustment function for the proportionality coefficient of the change in the radius of the protection area as follows:

[0074]

[0075] The function ensures that the value of k is not less than m, that is, the minimum radius of the protection area in the collision cone region during the conflict detection process is m times where 0 < m < 1 and 2m + n = 2. Compared with The optional speed range of each drone during flight is increased, ensuring that the drone flies more smoothly.

[0076] Step 5: Further optimize the flight speed of the drone to reduce the flight path length of the drone from the starting point to the target point. Conflict resolution is achieved and the flight path length of the unmanned aerial vehicle is shortened as much as possible.

[0077] Based on the gradient descent algorithm (Gradient Descent Optimization, GDO), optimize the flight path for conflict resolution of the drone by the reciprocal velocity obstacle method on the basis of Step 4.

[0078] There is a certain trade-off between optimizing the path length and avoiding collisions. Excessive pursuit of optimizing the path length may lead to an increased risk of collisions, while excessive focus on avoiding collisions may result in a longer path length.

[0079] After detecting a conflict, adjust the weights between the gradient descent algorithm and the RVO algorithm according to the actual scenario, and balance the two by introducing a certain collision cost into the optimized objective function. The overall objective function is in the following form:

[0080] J(P) = αL(P) + βC(P);

[0081] where J(P) is the overall optimization objective function, α and β are weight coefficients used to balance the path length and the collision cost, L(P) is the total length of the flight path of the drone at the current moment, C(P) is the collision cost function of the drone, and P is the position point of the drone, which changes with time and speed;

[0082] For one of the drones in the airspace, the optimized overall objective function is given as follows:

[0083]

[0084] Where N is the total flight time of the UAV from the starting point to the current position, λ is the weight of balancing the path length and collision cost, CollisionCost(P i ) is the collision cost function of the UAV at the i-th moment, P i is the flight position of the UAV at time i. The collision cost function of the I-th UAV at time i is as follows:

[0085]

[0086] Where M is the number of drone sorties in the airspace, P I and P J are the positions of the I-th UAV and the J-th UAV at the same flight time, D AB(min) The minimum safe distance between two drones is usually set to v I and v J are the speeds of the I-th UAV and the J-th UAV at the same flight time, v I and v J Determine by step 4-2, v max is the maximum flight speed of the UAV, ξ is the weight coefficient, and ξ evaluates the impact of the relative speed of two UAVs on the collision cost.

[0087] J(P i ) can be rewritten as:

[0088]

[0089] The above formula is the objective function to be optimized in the position domain. Fixing each flight time period of the drone as a constant, the corresponding objective function in the speed domain is obtained:

[0090]

[0091] According to the gradient descent algorithm, the flight speed is updated:

[0092]

[0093] in It represents solving the partial differential equation for the objective function, η is the speed update adjustment rate each time, and the value range of η is 0.001~0.01.

[0094] Since the learning rate is fixed, the process is always in the descent optimization process, but the flight path obtained cannot be guaranteed to be the shortest until the drone reaches the target point. To solve this problem, the decay coefficient is introduced so that the learning rate gradually decays during the gradient descent process, optimizing the flight speed and gradually generating the expected flight path:

[0095]

[0096] Here, i is the decay coefficient. As time goes by, the drone's flight path becomes longer and longer, and the learning rate value also becomes smaller and smaller, thereby ensuring that the speed update amplitude becomes slower and slower to obtain the flight speed that can obtain the approximate shortest path length. Until it reaches the target point and generates the final path.

[0097] The UAV uses the speed obtained by the above formula as the speed at the next flight moment to ensure that the flight path length is minimized to the maximum extent possible while resolving the conflict.

[0098] Furthermore, when it is determined in step 2-3 that no conflict resolution is required, the aircraft is flown to the target point along a straight course.

[0099] When the drone does not need to resolve conflicts, it will re-plan the optimal flight speed, direction and size based on its current position and the target point to fly towards the target point, preventing the drone from deviating from the planned route.

[0100] The beneficial effects of adopting the above technical solution are: the present invention provides a tactical conflict detection and resolution algorithm for low-altitude unmanned aerial vehicles in urban areas based on CCR-GDO-RVO, which considers the flight speed, direction and size of each drone to determine whether there is a flight conflict and whether emergency collision avoidance measures are required in the event of a flight conflict. The speed after considering the collision risk is secondary optimized to reduce the flight path length of each drone, thereby ensuring safer and more stable flight of the drone, improving flight efficiency and occupying less airspace resources.

[0101] The principle of the present invention is:

[0102] A drone equipped with an ADS-B receiver receives ADS-B messages. Message parsing reveals the drone's flight latitude, longitude, altitude, and speed. The drone's three-dimensional coordinate position relative to a fixed ground reference point is determined through Earth-centered, Earth-fixed, and East-North-Sky coordinate transformations. A spherical protection zone for the drone is constructed, along with a collision cone model. The drone's position and relative speed are used to determine whether a flight conflict exists. The collision cone model and flight speed are combined to determine a mutually beneficial speed barrier area. Conflict resolution is achieved by selecting a speed outside this area. Taking into account the collision risk, a regulation function is applied to adjust the radius of the drone's protection zone within the mutually beneficial speed barrier area in real time. This creates a collision-risk-constrained mutually beneficial speed barrier area. A drone's near-optimal flight speed outside this area is selected as the conflict resolution speed. The gradient descent algorithm is used to optimize the drone's flight conflict resolution speed within the previously calculated conflict resolution speed. If there is no flight conflict, the drone flies toward the target point at the calculated optimal speed, direction, and magnitude, preventing the drone from deviating from its intended route.

[0103] Compared with the prior art, the present invention has the following significant advantages:

[0104] According to the degree of collision risk, an adjustment function is assigned to adjust the radius of the drone protection zone in real time, expand the optional flight area of the drone, and then select a better flight speed, reducing the frequent flight jitter caused by the traditional speed barrier method and its variant (reciprocal speed barrier method). However, it cannot guarantee that the flight path for resolving flight conflicts is short, and does not consider problems such as path length optimization.

[0105] Based on the principle of gradient descent algorithm, the flight speed is further optimized. Under the premise of ensuring conflict resolution, the flight path of the drone is shortened and optimized to reduce the waste of airspace resources.

[0106] When the drone does not need to resolve conflicts, it will re-plan the optimal flight speed, direction and size based on its current position and the target point to fly towards the target point, preventing the drone from deviating from the planned route.

[0107] The beneficial effects of adopting the above technical solution are: the present invention provides a method for optimizing the path of tactical conflict detection and resolution of low-altitude unmanned aerial vehicles in urban areas. By considering the flight speed, direction and size of each drone, it is determined whether there is a flight conflict and whether emergency collision avoidance measures are required if there is a flight conflict. The speed after considering the collision risk is secondary optimized to reduce the flight path length of each drone, ensure that the drone flies more safely and smoothly, improve flight efficiency, and occupy less airspace resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0108] Figure 1 This is a flow chart of the CCR-GDO-RVO-based path optimization method for detecting and resolving tactical conflicts of low-altitude unmanned aerial vehicles in urban areas.

[0109] Figure 2 Schematic diagram of the Earth-centered Earth-fixed coordinate system of the present invention.

[0110] Figure 3 Schematic diagram of the flight status of two UAVs of the present invention.

[0111] Figure 4 Schematic diagram of the collision cone conflict detection model of the present invention.

[0112] Figure 5 Schematic diagram of the reciprocal speed barrier area model of the present invention.

[0113] Figure 6 Schematic diagram of collision risk assessment according to the present invention.

[0114] Figure 7 Schematic diagram of the reciprocal speed barrier area model considering collision risk of the present invention.

[0115] Figure 8 Schematic diagram of the relationship between the adjustment function curve and collision risk of the present invention.

[0116] Figure 9 Schematic diagram of the Sigmoid function curve of the present invention.

[0117] Figure 10 Schematic diagram of an example of the adjustment function curve of the present invention.

[0118] Figure 11 This is a schematic diagram of the flight state after gradient descent optimization of the present invention.

[0119] Figure 12 This is the flight path planning diagram of two UAVs after a conflict is detected according to the present invention. DETAILED DESCRIPTION

[0120] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, and the above and / or other advantages of the present invention will become more apparent.

[0121] In this embodiment, the urban low-altitude UAV tactical conflict detection and resolution algorithm based on CCR-GDO-RVO is as follows: Figure 1 As shown, the following steps are included:

[0122] Step 1: Each drone is equipped with a small Automatic Dependent Surveillance-Broadcast (ADS-B) receiver that continuously receives ADS-B messages and parses the ADS-B longitude and latitude coordinates in the form of Compact Position Reporting (CPR). This coordinate conversion is then performed in real time to obtain the drone's real-time relative ground position.

[0123] Step 1.1: Obtain ADS-B data message.

[0124] Get the drone position and speed data packets broadcast by ADS-B.

[0125] Step 1.2: Decode the ADS-B data.

[0126] The acquired binary ADS-B data message needs to be decoded for latitude, longitude and altitude. The altitude and speed can be obtained by converting the binary of the corresponding bits of the message into decimal. The latitude and longitude decoding method is CPR decoding, which converts the ADS-B data message into a readable text format.

[0127] In order to determine the exact location of the drone, the earth is first divided into many zones with different numbers, and then each zone is subdivided into many bins. The latitude zone position D is calculated. i :

[0128] D i =360° / (4N L -i)i=0,1;

[0129] Among them, i=0 is even coding, i=1 is odd coding, N L The number of latitude zones between the equator and the South Pole or the North Pole.

[0130] Calculate the latitude index j:

[0131]

[0132] Among them, l s is the latitude of the reference point, Y i is the CPR coordinate latitude, floor(x) represents the largest integer not greater than x, and MOD(a,b) represents the remainder after dividing a by b.

[0133] Calculate the latitude value R i :

[0134]

[0135] Determine the zone's east and west longitudes Li :

[0136]

[0137] Calculate the longitude index m:

[0138]

[0139] Among them, L s is the longitude of the reference point, X i is the CPR coordinate longitude.

[0140] Calculate the longitude value R Li for:

[0141]

[0142] Step 1-3: Convert the decoded ADS-B position data to coordinates, and convert the WGS-84 ellipsoid coordinates of the drone's ADS-B position and the reference position to Earth-centered Earth-fixed coordinates (X, Y, Z). Figure 2As shown, the East North Up (ENU) coordinates (x, y, z) of a drone A relative to a fixed reference point on the ground are calculated. The Earth-centered Earth-fixed coordinates of the drone are calculated as follows:

[0143]

[0144] Where H is the altitude of the drone. e is the eccentricity of the Earth, and N is the radius of curvature of the Earth. The calculation formulas for these two parameters are as follows:

[0145]

[0146] Among them, a is the semi-major axis of the Earth, and b is the semi-minor axis of the Earth.

[0147] The ground reference point coordinates are (0,0,0), and the corresponding WGS-84 ellipsoid coordinates are longitude (lon) and latitude (lat). The coordinates are calculated as follows:

[0148]

[0149] Similarly, the position coordinates of other unmanned aerial vehicles relative to the ground reference point can be obtained.

[0150] Step 2: Establish a flight protection zone for drones based on their flight speed and airspace building layout, and build a collision cone model for flight conflict detection.

[0151] Step 2-1: Determine the flight protection zone model and flight information of the UAV.

[0152] Define R 3 It is a three-dimensional real vector space, and the set of drone spherical protection zones V is as follows:

[0153]

[0154] Among them, d h Indicates the radius of the drone's spherical flight protection zone.

[0155] like Figure 3 As shown, suppose UAV A and UAV B are flying in a limited airspace. Through the ADS-B message analysis in step 1, the flight speeds of UAV A and UAV B are obtained as u1 and u2 respectively. With UAV A as the origin O as the ground fixed reference point, straight line AB as the x-axis, and the direction of vector AB as the positive direction of the x-axis, a three-dimensional space coordinate system is established by the right-hand rule.

[0156] In the three-dimensional spherical coordinate system, the flight speeds v1 and v2 of the two drones are obtained as follows:

[0157] v i =(u icosφ i sinθ i ,u i sinφ i sinθ i ,u i cosθ i ),i∈1,2;

[0158] Among them, φ i Represents the positive angle between the UAV's flight direction and the x-axis; θ i Represents the positive angle between the drone's flight direction and the z-axis.

[0159] The corresponding protection zone radii of UAV A and UAV B are and

[0160] Assume that the displacement generated by UAV A or UAV B during the kth time period Δt is Then the displacement of the UAV during the k+1th time period Δt is as follows:

[0161]

[0162] Define the maximum speed constraint v max and the maximum acceleration constraint a max Limit the available speed and acceleration of the drone as shown in the following formula:

[0163] ||v i ||≤v max ,i∈1,2;

[0164] ||a i ||≤a max ,i∈1,2;

[0165] Among them, ||.|| represents the vector modulus or Euclidean distance, v i is the speed of the i-th UAV, a i is the acceleration of the i-th UAV.

[0166] Step 2-2: Use the collision cone model to obtain the flight conflict detection area;

[0167] like Figure 4 As shown in the figure, taking two drones as an example, in three-dimensional space, the flight speed of drone A is v1, and the flight speed of drone B is v2. α is the angle between the line AB between the two drones and the boundary of the speed obstacle zone, that is, half of the collision cone vertex angle. γ is the angle between the line AB between the two drones and the relative speed v R The angle between them.

[0168] Define the relative collision cone (RCC) collision detection model, that is, the relative speed v at which drones may collide R = the velocity set of v1-v2, as shown in the following formula:

[0169]

[0170] in, is the relative velocity v R The ray drawn at the particle point of drone A; Sphere_B is the flight safety protection zone of drone B, and the collision cone boundary is tangent to Sphere_B.

[0171] Step 2-3: Determine whether a flight conflict occurs through the collision cone model;

[0172] The collision cone model only considers the positional relationship and current state between drones. With drone B as the reference point, drone A moves relative to drone B. If the speed of drone A relative to drone B is within the collision cone, a flight conflict will occur between the two drones; otherwise, there will be no flight conflict. Based on the collision cone model, the following geometric judgment method is used to detect the flight conflict between two drones: when α>γ, there is a flight conflict between the two drones; otherwise, there is no flight conflict. The expressions of α and γ are shown in the following formulas:

[0173]

[0174] Among them D AB is the relative distance between UAV A and UAV B, D AB is vector AB.

[0175] Step 3: Use the Reciprocal Velocity Obstacle (RVO) conflict detection algorithm to determine the conflict area, assign the UAV a flight speed, and resolve the UAV flight conflict.

[0176] Step 3-1: Obtain the reciprocal speed barrier area through the collision cone area and the flight speed of the drone.

[0177] Still taking two UAVs as an example, the traditional speed barrier method is iterated bidirectionally to obtain the reciprocal speed barrier method. This algorithm attempts to simultaneously consider the reciprocal motion of UAV A and UAV B, so that the two UAVs share the responsibility of obstacle avoidance to produce a smoother avoidance path. The RVO area generated by A relative to B can be obtained as follows: Figure 5 As shown, the region generation expression is as follows:

[0178]

[0179] in, represents the Minkowski sum; the next moment's flight speed v of the drone A in the area satisfies the following formula:

[0180]

[0181] Where t is a certain flight time in the future, and Sphere_B is the flight safety protection zone of UAV B.

[0182] If there are n drones flying in the airspace, the flight conflict resolution speed set for drone A can be expressed as follows:

[0183]

[0184] Where n is the number of drones flying in the airspace.

[0185] Step 3-2: Assign the drone a flight speed to resolve drone flight conflicts.

[0186] Set an optimal flight speed for drone A As shown in the following formula:

[0187]

[0188] in, Indicates the optimal flight speed of drone A, P1 indicates the current position of drone A, Indicates the final position of UAV A;

[0189] Drone A selects a speed vector closest to the optimal flight speed outside the RVO area as the new flight speed As shown in the following formula:

[0190]

[0191] Among them, v is the flight speed set of UAV A at the next moment.

[0192] Step 4: Based on the RVO algorithm, the conflict detection algorithm of Consider Collision Risk-Reciprocal Velocity Obstacle (CCR-RVO) that further considers the collision risk is used to reduce the conflict area and adjust the UAV flight conflict resolution speed.

[0193] Step 4-1: Consider the collision risk of the two drones based on their relative positions and speeds, and select an appropriate protection zone radius for the drones in real time.

[0194] Increase the range of the optional obstacle avoidance speed vector of the UAV at each moment according to the real-time change of the protection area radius. The degree of collision risk between UAVs can be evaluated by the gradient color collision cone area as shown in Figure 6 The color of the area gradually changes from white to gray and then to black as the distance between UAV A and UAV B decreases. At this time, the degree of collision risk for the two UAVs gradually increases, and the possibility of collision also increases.

[0195] Set a proportionality coefficient k for the change of the protection area radius, where 0 < k ≤ 1, so that the protection area radius in the traditional collision cone can be reduced at the next flight moment, thereby reducing the RCC. The generated RVO area is as shown in Figure 7 This is the reciprocal velocity obstacle area considering collision risk. By reducing the protection area radius in the traditional collision cone to reduce the RCC area, the generated RVO area is the reciprocal velocity obstacle area considering collision risk; thus increasing the speed area available for the UAV to choose.

[0196] If the relative distance between UAVs is relatively close, expand the protection area radius at the next flight moment, thereby expanding the RCC area and reducing the speed area available for the UAV to choose; when the relative distance decreases and approaches the protection area radius, the protection area radius remains unchanged.

[0197] Step 4-2: Adjust the flight conflict resolution speed of the UAV;

[0198] UAV A expands the range of the optional flight conflict resolution speed set. Then select a better flight conflict resolution speed to reduce the problem of frequent jitter caused by the overly large speed obstacle area of the traditional algorithm and a relatively large number of UAVs in the airspace. The optimal flight speed The calculation formula is as shown below:

[0199]

[0200] In this paper, a speed vector closest to the optimal flight speed outside the CCR-RVO area is selected as the new flight speed for UAV A, as shown below:

[0201]

[0202] Similarly, the flight conflict resolution speed of UAV B is obtained.

[0203] When multiple UAVs are flying, the flight conflict resolution speed of UAV A in step 4-2 is calculated by the following formula:

[0204]

[0205] where n is the number of UAVs flying in the airspace.

[0206] Similarly, the flight conflict resolution speed of UAV B can also be obtained.

[0207] The specific method for setting the protection area radius in step 4 is as follows:

[0208] Select an adjustment function for the protection area radius change ratio coefficient. The independent variable of this function is the distance between the two UAVs, and the dependent variable is the ratio coefficient. It is stipulated that the smaller the distance between the UAV and other UAVs or obstacles, the larger the k value, and as the collision risk increases, the growth rate of the k value gradually increases, that is, the slope of the adjustment function at k gradually increases. On the contrary, the larger the distance, the smaller the k value, and the decreasing rate of the k value gradually decreases, that is, the slope of the adjustment function at k gradually decreases;

[0209] The relationship between the adjustment function curve and the collision risk is as Figure 8 shown.

[0210] The Sigmoid function has the property that the function value increases as the independent variable increases, and when the independent variable increases, the slope at the corresponding function point also gradually increases, and the slope change of the points on the function curve is continuous. The expression of the Sigmoid function is shown as follows:

[0211]

[0212] Its function curve is as Figure 9 shown.

[0213] Based on the characteristics of the Sigmoid function, select an adjustment function as shown in the following formula to ensure that the change process of the CCR - RVO area is also continuous;

[0214]

[0215] The function ensures that the k value is not less than m, that is, the minimum protection area radius of the collision cone area during the conflict detection process is m times where 0 < m < 1 and 2m + n = 2. Compared with the selectable speed range of each UAV during flight is increased, ensuring that the UAV flies more smoothly.

[0216] As Figure 10 shown, a sample function curve is given, and the corresponding adjustment function is as follows:

[0217]

[0218] Step 5: Further optimize the flight speed of the UAV to reduce the flight path length of the UAV from the starting point to the target point.

[0219] Gradient Descent Optimization (GDO) is an iterative method for optimizing an objective function, widely used in machine learning, deep learning, and optimization problems. Its core idea is to find a local or global minimum by continuously adjusting parameters to gradually reduce the value of the objective function.

[0220] According to the principle of Gradient Descent Optimization (GDO), based on step 4, the reciprocal speed barrier method is used to optimize the flight path of UAV conflict resolution.

[0221] In practical applications, optimizing flight path length is naturally important, but there is a trade-off between optimizing path length and collision avoidance. Over-focusing on path length optimization may increase the risk of collisions, while over-focusing on collision avoidance may result in longer paths. Therefore, after detecting a collision, the weights between the gradient descent algorithm and the RVO algorithm need to be adjusted based on the actual scenario. A certain collision cost needs to be introduced into the optimization objective function to balance the relationship between the two. The overall objective function should have the following form:

[0222] J(P)=αL(P)+βC(P);

[0223] Among them, J(P) is the overall optimization objective function, α and β are weight coefficients, which are used to balance the path length and collision cost, L(P) is the total length of the UAV flight path at the current moment, C(P) is the collision cost function of the UAV, and P is the UAV position point, which changes with time and speed.

[0224] For one of the drones in the airspace, the overall objective function to be optimized is given as follows:

[0225]

[0226] Where N is the total flight time of the UAV from the starting point to the current position, λ is the weight of balancing the path length and the collision cost, CollisionCost(P i ) is the collision cost function of the UAV at the i-th moment, P i is the flight position of the UAV at time i. The collision cost function of the I-th UAV at time i is as follows:

[0227]

[0228] Where M is the number of drone sorties in the airspace, P I and P J are the positions of the I-th UAV and the J-th UAV at the same flight time, D AB(min)The minimum safe distance between two drones is usually set to v I and v J are the speeds of the I-th UAV and the J-th UAV at the same flight time, v I and v J Determine by step 4-2, v max is the maximum flight speed of the UAV, ξ is the weight coefficient, and ξ evaluates the impact of the relative speed of two UAVs on the collision cost.

[0229] J(P i ) can be rewritten as:

[0230]

[0231] The above formula is the objective function to be optimized in the position domain. Fixing each flight time period of the drone as a constant, the corresponding objective function in the speed domain is obtained:

[0232]

[0233] According to the gradient descent algorithm, the flight speed is updated:

[0234]

[0235] in It represents solving the partial differential equation for the objective function, η is the speed update adjustment rate each time, and the value range of η is 0.001~0.01.

[0236] Since the learning rate is fixed, the process is always in the descent optimization process, but the flight path obtained cannot be guaranteed to be the shortest until the drone reaches the target point. To solve this problem, the decay coefficient is introduced so that the learning rate gradually decays during the gradient descent process, optimizing the flight speed and gradually generating the expected flight path:

[0237]

[0238] Here, i is the decay coefficient. As time goes by, the drone's flight path becomes longer and longer, and the learning rate value also becomes smaller and smaller, thereby ensuring that the speed update amplitude becomes slower and slower to obtain a flight speed that can obtain the approximate shortest path length. Until it reaches the target point and generates the final path.

[0239] The UAV takes the speed obtained by the above formula as the speed of the next flight moment to ensure that the flight path length is minimized as much as possible under the premise of conflict resolution, so that the speed gradually deviates to the target point. The optimized UAV flight state is as follows: Figure 11 As shown, the optimized speed is This allows the drone to deviate as much as possible along the direction of the optimal flight speed based on conflict resolution, thus minimizing the flight path length.

[0240] When no conflict resolution is required, fly to the target point along a straight course.

[0241] When the drone does not need to resolve conflicts, it will re-plan the optimal flight speed, direction and size based on its current position and the target point to fly towards the target point, preventing the drone from deviating from the planned route.

[0242] like Figure 12 As shown in the figure, when a flight conflict is detected, UAV A's flight path will pass through points M and N. MN is the path flown during the avoidance process, and the velocity after the avoidance points to the corresponding target point. UAV B's flight path will pass through points P and Q. PQ is the path flown during the avoidance process, and the velocity after the avoidance points to the corresponding target point.

[0243] The present invention provides a method and approach for optimizing the path for detecting and resolving tactical conflicts between low-altitude unmanned aerial vehicles (UAVs) in urban areas. While there are numerous methods and approaches for implementing this technical solution, the foregoing merely represents a preferred embodiment of the present invention. It should be noted that those skilled in the art may make improvements and modifications without departing from the principles of the present invention, and such improvements and modifications are also within the scope of protection of the present invention. Any components not specified in this embodiment may be implemented using existing technologies.

Claims

1. A method for optimizing the path of tactical conflict detection and resolution for low-altitude unmanned aerial vehicles in urban areas, characterized by: The following steps are involved: Step 1: The drone continuously receives Automatic Dependent Surveillance-Broadcast (ADS-B) messages through its receiver, parses the ADS-B longitude and latitude coordinates through the Concise Position Report (CPR), and performs real-time coordinate conversion to obtain the drone's real-time relative ground position. Step 2: Based on the UAV flight speed and airspace building layout obtained in step 1, establish the UAV flight protection zone and build a collision cone model for flight conflict detection; Step 3: Based on the flight conflict detection results in step 2, the RVO conflict detection algorithm is used to determine the conflict area, and the flight conflict is resolved by assigning the UAV a flight speed. Step 4: Based on the RVO algorithm, the conflict area obtained in step 3 is reduced by using the CCR-RVO conflict detection algorithm. A flight safety protection zone radius adjustment function is set to adjust the radius of the protection zone. The flight speed of the UAV is adjusted according to the adjusted protection zone radius. Step 5: Reduce the length of the UAV's flight path from the starting point to the target point by optimizing the UAV's flight speed in step 4, resolve the conflict and shorten the UAV's flight path as much as possible.

2. The method for optimizing the path of tactical conflict detection and resolution of low-altitude unmanned aerial vehicles in urban areas according to claim 1 is characterized in that: The specific method of step 1 is: Step 1-1: Obtain ADS-B data message, the data including the drone's position and speed; Step 1-2: Decode the ADS-B data. Decode the longitude, latitude, altitude, and speed of the UAV in the binary ADS-B data message obtained in step 1-1. Convert the corresponding bits of the message to decimal to obtain the altitude and speed information. Decode the longitude and latitude information by decoding the CPR of the message. Step 1-3: Perform coordinate conversion on the decoded ADS-B position data obtained in step 1-2, convert the WGS-84 ellipsoid coordinates of the drone's ADS-B position and the reference position into Earth-centered Earth-fixed coordinates (X, Y, Z), and calculate the east-north-sky coordinates (x, y, z) of the drone relative to the ground fixed reference point using the Earth-centered Earth-fixed coordinates (X, Y, Z).

3. The method for optimizing the path of tactical conflict detection and resolution of low-altitude unmanned aerial vehicles in urban areas according to claim 2 is characterized in that: The specific method of step 2 is: Step 2-1: Establish a UAV flight protection zone model and determine flight information; Define R 3 It is a three-dimensional real vector space, and the set of drone spherical protection zones V is as follows: Among them, d h Indicates the radius of the drone's spherical flight protection zone; Assume that drones A and B are flying in a limited airspace. Through the ADS-B message analysis in step 1, the flight speeds of drones A and B are obtained as u1 and u2 respectively. With drone A as the origin O as the ground fixed reference point, line AB as the x-axis, and the direction of vector AB as the positive direction of the x-axis, a three-dimensional space coordinate system is established using the right-hand rule. In the three-dimensional spherical coordinate system, the flight speeds v1 and v2 of the two drones are obtained as shown in the following formula: v i =(u i cosφ i sinθ i ,u i sinφ i sinθ i ,u i cosθ i ),i∈1,2; Among them, φ i Represents the positive angle between the UAV's flight direction and the x-axis; θ i Represents the positive angle between the drone's flight direction and the z-axis; The corresponding protection zone radii of UAV A and UAV B are and Assume that the displacement generated by UAV A or UAV B during the kth time period Δt is Then the displacement of the UAV during the k+1th time period Δt is as follows: Define the maximum speed constraint v max and the maximum acceleration constraint a max Limit the available speed and acceleration of the drone as shown in the following formula: ||in i ||≤v max ,i∈1,2; ||a i ||≤a max ,i∈1,2; Among them, ||.|| represents the vector modulus or Euclidean distance, v i is the speed of the i-th UAV, a i is the acceleration of the i-th UAV; Step 2-2: Use the collision cone model to obtain the flight conflict detection area; In three-dimensional space, the flight speed of UAV A is v1, the flight speed of UAV B is v2, α is the angle between the line AB between the two UAVs and the boundary of the speed obstacle zone, that is, half of the collision cone vertex angle, γ is the angle between the line AB between the two UAVs and the relative speed v R The angle between Define the relative collision cone RCC model, that is, the relative speed v at which drones may collide R = the velocity set of v1-v2, as shown in the following formula: in, is the relative velocity v R The ray drawn at the mass point of drone A; Sphere_B is the flight safety protection zone of drone B, and the collision cone boundary is tangent to Sphere_B; Step 2-3: Determine whether a flight conflict occurs through the collision cone model; The collision cone model considers the positional relationship and current state of the two drones. With drone B as the reference point, drone A moves relative to drone B. If the speed of drone A relative to drone B is within the collision cone, a flight collision will occur between the two drones; otherwise, there will be no flight conflict. The geometric judgment method for detecting flight conflicts between two UAVs based on the collision cone model is as follows: when α > γ, there is a flight conflict between the two UAVs; otherwise, there is no flight conflict. The expressions for α and γ are shown in the following formula: Among them D AB is the relative distance between UAV A and UAV B, D AB is vector AB.

4. The method for optimizing the path for detecting and resolving tactical conflicts of low-altitude unmanned aerial vehicles in urban areas according to claim 3 is characterized in that: When it is judged in step 2-3 that conflict resolution is not required, the UAV flies straight towards the target point along the heading.

5. The method for optimizing the path of tactical conflict detection and resolution of low-altitude unmanned aerial vehicles in urban areas according to claim 4 is characterized in that: The specific method of step 3 is as follows: Step 3-1: Obtain the reciprocal velocity obstacle region through the collision cone region and the flight speed of the UAV. The reciprocal velocity obstacle conflict detection algorithm simultaneously considers the reciprocal motion of UAV A and UAV B, enabling the two UAVs to jointly bear the responsibility of obstacle avoidance. The expression for the RVO region generated by A relative to B can be obtained: in, represents the Minkowski sum; the next moment's flight speed v of the drone A in the area satisfies the following formula: where t is a future flight moment, and Sphere_B is the flight safety protection area of UAV B. Step 3-2: Assign a flyable speed to the UAV to achieve the resolution of the UAV flight conflict. Set an optimal flight speed for drone A As shown in the following formula: in, Indicates the optimal flight speed of drone A, P1 indicates the current position of drone A, Indicates the final position of UAV A; Drone A selects a speed vector closest to the optimal flight speed outside the RVO area as the new flight speed As shown in the following formula: where v is the set of flight speeds of UAV A at the next moment.

6. The method for optimizing the path for detecting and resolving tactical conflicts of low-altitude unmanned aerial vehicles in urban areas according to claim 5 is characterized in that: If there are n UAVs flying in the airspace, the set of flight conflict resolution speeds of UAV A in step 3-1 is calculated by the following formula: where n is the number of UAVs flying in the airspace.

7. The method for optimizing the path of tactical conflict detection and resolution of low-altitude unmanned aerial vehicles in urban areas according to claim 6 is characterized in that: The specific method of step 4 is as follows: Step 4-1: Consider the collision risk between the two UAVs based on their relative positions and speeds, and select a protection area radius for the UAVs in real time. Increase the range of the optional obstacle avoidance speed vectors for the UAVs at each moment according to the real-time change of the protection area radius. Judge the degree of collision risk between the UAVs and adjust the RCC region. Specifically: When the relative distance between the UAVs is far, reduce the protection area radius in the traditional collision cone to reduce the RCC region. At this time, the generated RVO region is the reciprocal velocity obstacle region considering the collision risk; thereby increasing the speed region available for the UAVs. When the relative distance of the UAV is relatively close, the radius of the protection zone at the next flight moment is expanded, thereby expanding the RCC area and reducing the speed range available for the UAV; when the relative distance decreases and approaches the protection zone radius, the protection zone radius remains remain unchanged; When UAV A expands the flight conflict resolution speed set area, an optimal flight speed is set for UAV A. As shown in the following formula: Step 4-2: Adjust the flight conflict resolution speed of the UAVs. UAV A selects a speed vector that is closest to the optimal flight speed outside the CCR-RVO region as the new flight speed of UAV A, as shown in the following formula:

8. The method for optimizing the path for detecting and resolving tactical conflicts of low-altitude unmanned aerial vehicles in urban areas according to claim 7 is characterized in that: Similarly, the flight conflict resolution speed of UAV B is obtained. When multiple UAVs are flying, the flight conflict resolution speed of UAV A in step 4-2 is calculated by the following formula:

9. The method for optimizing the path for detecting and resolving tactical conflicts of low-altitude unmanned aerial vehicles in urban areas according to claim 8 is characterized in that: where n is the number of UAVs flying in the airspace. The specific method of setting the protection area radius in step 4 is as follows: Set the change ratio coefficient k of the protection area radius, and 0 < k ≤ 1. Based on the requirements of the ratio coefficient k and the characteristics of the Sigmoid function, select the adjustment function of the change ratio coefficient of the protection area radius, as shown below:

10. The method for optimizing the path of tactical conflict detection and resolution of low-altitude unmanned aerial vehicles in urban areas according to claim 9 is characterized in that: where, 0 < m < 1, 2m + n = 2. The specific method of step 5 is as follows: Optimize the flight path for UAV conflict resolution using the reciprocal velocity obstacle method based on the gradient descent algorithm GDO in step 4. After detecting a conflict, adjust the weight between the gradient descent algorithm and the RVO algorithm according to the actual scenario, and balance the two by introducing a certain collision cost into the optimized objective function. The overall objective function is in the following form: J(P) = αL(P) + βC(P); Among them, J(P) is the overall optimization objective function, α and β are weight coefficients used to balance the path length and collision cost, L(P) is the total length of the UAV flight path at the current moment, C(P) is the collision cost function of the UAV, and P is the UAV position point, which changes with time and speed; For one of the drones in the airspace, the overall objective function to be optimized is as follows: Where N is the total flight time of the UAV from the starting point to the current position, λ is the weight of balancing the path length and collision cost, CollisionCost(P i ) is the collision cost function of the UAV at the i-th moment, P i is the flight position of the UAV at the i-th moment, where the collision cost function of the I-th UAV at the i-th moment is as follows: Where M is the number of drone sorties in the airspace, P I and P J are the positions of the I-th UAV and the H-th UAV at the same flight time, D AB(min) The minimum safe distance between two drones is usually set to v I and v J are the speeds of the I-th UAV and the J-th UAV at the same flight time, v I and v J Determine by step 4-2, v max is the maximum flight speed of the UAV, ξ is the weight coefficient, and ξ evaluates the influence of the relative speed of the two UAVs on the collision cost; J(P i ) can be rewritten as: The above formula is the objective function to be optimized in the position domain. Fixing each flight time period of the drone as a constant, the corresponding objective function in the speed domain is obtained: According to the gradient descent algorithm, the flight speed is updated: in Represents solving the partial differential equation for the objective function, η is the speed update adjustment rate each time, and the value range of η is 0.001~0.01; By introducing the attenuation coefficient in the gradient descent process, the problem of the flight path always being in the descent optimization process is solved. The flight path length is minimized as much as possible under the premise of conflict resolution, and the expected flight path is generated as follows: Where i is the attenuation coefficient, which increases with time.

Images