A Quadrotor UAV Swarm Obstacle Avoidance Method Based on Vision-of-View and Speed ​​Guidance

Abstract

This invention discloses a quadrotor UAV swarm obstacle avoidance method based on the field-of-view (LOR) method and velocity guidance. The method designs an adaptive communication topology mechanism for the swarm based on the LLO method; constructs smooth communication weight coefficients based on the LLO's field-of-view angle and relative distance; designs collision avoidance and styling maintenance control inputs within the UAV swarm based on the "far attracts near repulses" principle; designs the obstacle avoidance guidance velocity term based on the limit cycle method; and constructs the UAV swarm obstacle avoidance control inputs by combining the artificial potential field method. This method enables the UAV swarm to safely navigate in complex obstacle environments while maintaining its swarm morphology, shortens obstacle avoidance time, and improves obstacle avoidance efficiency.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) swarms, specifically to a method for obstacle avoidance in quadrotor UAV swarms based on vision-based and speed-guided approaches. Background Technology

[0002] In the face of future urban warfare characterized by information scarcity and constant change, drone swarms, with their low cost and large-scale swarm configuration, will compensate for the limited operational range and high mission failure rate of individual drones, becoming an important means of air superiority. Flexible and efficient obstacle avoidance flight, however, becomes a key technical challenge in ensuring the safety of drone swarms.

[0003] Based on the information exchange methods among swarm drones, swarms can be divided into centralized and distributed types. Centralized swarms have a simple obstacle avoidance principle and are easy to optimize globally, but they require high communication bandwidth and are not suitable for combat environments. In a distributed swarm, each drone can be considered a relatively independent intelligent agent. Drones in the swarm interact with their neighbors through a local communication network, making decisions collaboratively and independently.

[0004] When drone swarms attempt obstacle avoidance, they often need to disrupt the swarm formation. This can lead to a difficult decision-making process where drones struggle between maintaining the swarm formation and obstacle avoidance, resulting in issues such as hovering and stalling. The design and coordination of individual obstacle avoidance mechanisms and swarm maintenance mechanisms for swarm drones is a crucial research area in swarm obstacle avoidance. Summary of the Invention

[0005] Purpose of the invention: The purpose of this invention is to enable unmanned aerial vehicle (UAV) swarms to safely navigate through complex obstacle environments while maintaining their swarm formation.

[0006] Technical solution:

[0007] A method for obstacle avoidance in a quadcopter drone swarm based on vision-of-view and speed-guided approaches includes the following steps:

[0008] Step 1: Based on the UAV motion model, design the UAV control input, which includes UAV swarm behavior motion control terms;

[0009] Step 2: Design an adaptive communication topology mechanism for UAV swarms based on the field-of-view method;

[0010] Step 3: Construct smooth adaptive communication weight coefficients based on the neighbor's field of view and relative distance;

[0011] Step 4: Based on the principle of "repulsion from afar and attraction from afar," represent the motion control terms for the drone swarm behavior;

[0012] Step 5: Design the obstacle avoidance guidance velocity term based on the limit cycle method, and construct the external obstacle avoidance control input of the UAV swarm using the artificial potential field method.

[0013] Preferably, the UAV motion model in step 1 is:

[0014]

[0015] Where, p i =[p ix ,p iy ] T Let v be the two-dimensional spatial position vector of UAV i, with the superscript "·" representing the derivative. i =[v ix ,v iy ] T Let u be the two-dimensional spatial velocity vector of UAV i. i =[u ix ,u iy ] T Let x and y be the control inputs for UAV i, and let T be the transpose of the two-dimensional space x and y axes, respectively. The velocity of the UAV motion model satisfies the following relationship with the control inputs:

[0016]

[0017]

[0018] Wherein, "||||" represents taking the L2 norm, and the control input u is determined according to the UAV motion model. i Represented as:

[0019]

[0020] in For the motion control of drone swarm behavior, To avoid external obstacles.

[0021] Preferably, the implementation process of step 2 is as follows:

[0022] First, calculate the angle θ between the direction of motion of UAV j and UAV i. ij :

[0023]

[0024] Where p j =[p jx ,p jy ] T Let be the two-dimensional spatial position vector of UAV j, and the symbol "·" represents the dot product;

[0025] Next, determine the included angle θ. ijIs it within the drone's field of view (ω)?

[0026] |θ ij |≤|ω|

[0027] Based on the drone's speed j, the angle ω spreads outwards to both sides.

[0028] Determine if the distance between drone j and drone i is less than the communication distance r:

[0029] ||p j -p i ||<r

[0030] There is an adaptive communication topology mechanism for drone swarms. for:

[0031]

[0032] Preferably, step 3 is implemented by selecting a smoothing function ρ. h (z):

[0033]

[0034] In the formula, h is the smoothing coefficient; z is a parameter that determines the degree of smoothing.

[0035] Constructing smooth adaptive communication weight coefficients a ij :

[0036] a ij =s ij (θ)d ij (p)

[0037] Among them, s ij (θ ij )=ρ h (θ ij / ω) is the field-of-view weighting coefficient, indicating that the closer an individual is to the extended line of the body's velocity, the greater the influence of the body; d ij (p)=ρ h (z)(||p i -p j || / r) is the distance weighting coefficient, which means that the closer an individual is to the host, the greater the influence of the host. The host and the individual refer to UAV i and UAV j, respectively.

[0038] Preferably, in step 4, the drone swarm behavior motion control term is divided into two parts, namely:

[0039]

[0040] in, This serves as the input for coordinated collision avoidance speed control within a drone swarm. The attraction of the target point to the drone swarm and the ideal speed of the drone swarm;

[0041]

[0042] in, and All are adjustable parameters;

[0043] Let φ(·) represent the unit vector pointing from UAV i to UAV j, and let φ(·) be the power function.

[0044]

[0045] Where a, b, and c are all parameters;

[0046]

[0047] Where, p γ v represents the target point location. γ For the ideal movement speed of the drone swarm, meet and These are all parameter adjustments; all drones in the drone swarm are aligned with the target point and the ideal speed.

[0048] The motion control term for drone swarm behavior is then expressed as:

[0049]

[0050] Preferably, step 5 is implemented as follows: First, the obstacle is made convex, the minimum circumcircle of the obstacle is drawn, and the obstacle threat radius r is set. s The drone swarm enters the obstacle threat radius r s When inside, it will be subject to the repulsive force of obstacles. Based on the drone's position information, a repulsive force function φ is constructed. β Find the repulsive force on the drone:

[0051]

[0052] In the formula, β refers to the coefficient related to obstacle avoidance. Let i be the closest position of drone i to the k-th obstacle, with weight coefficients. The range of the repulsive force is smoothed. For the obstacle avoidance vector direction, Here is the expression for the repulsive force function;

[0053] Establish a coordinate system with the center of the obstacle as the origin to describe the position and velocity information of the UAV. Obstacle coordinate system O xy With the geodetic coordinate system Gxy The conversion relationship is as follows:

[0054] O xy =G xy -[x o ,y o ] T

[0055] In the formula, [x o ,y o ] T The coordinates of the obstacle in the geodetic coordinate system;

[0056] Based on the velocity direction of the UAV, a linear equation is constructed and simplified to the standard linear form, as follows:

[0057] ax + by + c = 0

[0058] Where a, b, and c are parameters;

[0059] The formula for calculating the closest distance d between line l and the center of the circle is as follows:

[0060]

[0061] Substituting d into the limit cycle equation, we obtain the direction of the guiding velocity.

[0062]

[0063]

[0064] In the formula, the asymptotic factor α = 1 / r s 2 The distance d determines the direction of obstacle avoidance. When d is positive, the limit cycle rotates clockwise; when d is negative, the limit cycle rotates counterclockwise. The radius of convergence of the limit cycle is set to r. s ;

[0065] Transform the ideal velocity v0 to the geodetic coordinate system, and rotate the velocity v of the UAV i. i By rotating to the ideal velocity v0 direction, we can obtain the guiding velocity of the i-th drone under the influence of the k-th obstacle.

[0066]

[0067] The external obstacle avoidance control input for the drone swarm is:

[0068]

[0069] in, and These are all adjustable parameters.

[0070] Beneficial effects:

[0071] 1. The present invention provides a collision avoidance method for quadrotor UAV swarms based on vision-based and speed-guided methods. It adopts vision-based design of swarm adaptive communication topology mechanism and combines the "far-attracting and near-repelling" force principle to solve the difficulties of swarm separation and swarm maintenance.

[0072] 2. The present invention provides a collision avoidance method for quadrotor UAV swarms based on vision-of-view and speed-guided methods. The method introduces the concept of weight coefficients in both swarm maintenance and obstacle avoidance, and uses a smooth function to smooth the vision-of-view boundary and distance boundary to prevent jitter.

[0073] 3. The present invention provides a collision avoidance method for quadrotor UAV swarms based on the vision method and velocity guidance. It combines the artificial potential field method with the limit cycle method to construct the obstacle avoidance guidance velocity term, which solves the problems of swarm hovering and stagnation during obstacle avoidance, and finally enables the swarm to pass through the obstacle area quickly and smoothly. Attached Figure Description

[0074] Figure 1 This is a diagram of the cluster individual obstacle avoidance algorithm structure;

[0075] Figure 2 This is a schematic diagram of the field of view of an individual drone;

[0076] Figure 3 It is the graph of a smooth function;

[0077] Figure 4 This is a schematic diagram of the minimum circumcircle of the obstacle;

[0078] Figure 5 This is a schematic diagram of a limit cycle;

[0079] Figure 6 This is the obstacle avoidance trajectory diagram of the drone swarm in scenario 1;

[0080] Figure 7 This is the obstacle avoidance trajectory diagram of the drone swarm in scenario 2;

[0081] Figure 8 This is the obstacle avoidance trajectory diagram of the drone swarm in scenario 3;

[0082] Figure 9 This is a graph showing the changing trend of the distance between adjacent drones in scenario 3.

[0083] Figure 10 This is a graph showing the trend of the drone's x-axis velocity in scenario 3.

[0084] Figure 11 This is a graph showing the trend of the drone's y-axis velocity in scenario 3. Detailed Implementation

[0085] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0086] The specific design steps of the obstacle avoidance method for quadrotor UAV swarms based on vision-of-view and velocity guidance are as follows:

[0087] Step 1: Based on the UAV motion model, design an overall obstacle avoidance control system for the UAV swarm, including UAV swarm behavior motion control items and obstacle avoidance control items.

[0088] In step 1, the simplified motion model of the UAV is introduced:

[0089]

[0090] In the formula, p i =[p ix ,p iy ] T Let v be the two-dimensional spatial position vector of UAV i. i =[v ix ,v iy ] T Let u be the two-dimensional spatial velocity vector of UAV i. i =[u ix ,u iy ] T This is the control input for drone i. The velocity and acceleration of this model satisfy:

[0091]

[0092]

[0093] Control methods are designed based on the motion model of the unmanned aerial vehicle (UAV):

[0094]

[0095] The control input consists of two superimposed parts, in which For the motion control of drone swarm behavior, To avoid external obstacles, the obstacle avoidance algorithm structure diagram is as follows: Figure 1 As shown.

[0096] Step 2: Design a cluster adaptive communication topology mechanism based on the vision method.

[0097] In step 2, firstly, the angle θ between the motion direction of UAV j and UAV i is calculated. ij :

[0098]

[0099] Determine whether the included angle θ falls within the range of ω:

[0100] |θ ij |≤|ω|

[0101] In the formula, ω represents the UAV's field of view, which expands outwards from the individual's speed by an angle of ω. Figure 2 As shown.

[0102] Determine if the distance between drone j and drone i is less than the communication distance r:

[0103] ||p j -p i ||<r

[0104] The neighbor set is:

[0105]

[0106] Step 3: Construct smooth communication weight coefficients based on the neighbor's field of vision and relative distance.

[0107] In step 3, the smoothing function ρ is selected. h (z):

[0108]

[0109] In the formula, h is the smoothing coefficient, which determines the degree of smoothing. The graph of the function is as follows: Figure 3 As shown.

[0110] Construct smooth weighting coefficients a ij :

[0111] a ij =s ij (θ)d ij (p)

[0112] In the formula, s ij (θ)=ρ h (θ / ω) is the field-of-view weighting coefficient, indicating that the closer an individual is to its neighbors along the extension of its velocity, the greater the influence they have on that individual. ij (p)=ρ h (||p i -p j || / r) represents the distance weighting coefficient; similarly, the closer the neighbors are, the greater the influence.

[0113] Step 4: Based on the principle of "repelling from afar and attracting from afar", design the drone swarm configuration to maintain control input.

[0114] In step 4, the cluster motion control input is divided into two parts:

[0115]

[0116] In the formula, This serves as the input for coordinated collision avoidance speed control within a drone swarm. The attraction of the target point to the cluster and the ideal speed of the cluster.

[0117]

[0118] In the formula, φ represents the unit vector pointing from drone i to drone j, and φ is the power function. Drones follow the "near-repulsion, far-attraction" principle of force to ensure a collision-free and orderly structure within the cluster, achieving speed matching through consistency.

[0119]

[0120] In the formula, p γ v represents the target point location. γ For the ideal movement speed of the cluster, satisfy All drones in the cluster are aiming at the target point and the ideal speed.

[0121] Step 5: Design the obstacle avoidance guidance velocity term based on the limit cycle method, and construct the external obstacle avoidance control input of the cluster by combining the artificial potential field method.

[0122] In step 5, firstly, the obstacle is made convex by drawing a minimum circumcircle such as... Figure 4 As shown, an obstacle threat radius is set. When the swarm of drones enters the obstacle threat range, it will be subject to the repulsive force of the obstacle. Based on the drone's position information, a repulsive force function φ is constructed. β Find the repulsive force on the drone:

[0123]

[0124] In the formula, The weight coefficient represents the closest position of the drone to the k-th obstacle. The range of the repulsive force is smoothed. For the obstacle avoidance vector direction, This is the expression for the repulsive force function. Obstacle avoidance using only position information and an artificial potential field method is prone to getting stuck in local extrema and its effectiveness is slow. Therefore, a velocity guidance term is needed to help the drone escape obstacles.

[0125] Step 5 also includes constructing a limit cycle at the center of the obstacle to guide the individual out of the local stagnation or wandering region (local extreme point). The limit cycle ensures that all points within its domain converge to the unit circle, such as... Figure 5As shown, the boundary of obstacle action can be flexibly designed. The collision threat radius of the obstacle is used as the convergence radius of the limit cycle. The obstacle avoidance direction, i.e., the rotation direction of the limit cycle, is selected according to the position of the individual's velocity direction relative to the center of the obstacle. The limit cycle vector direction at each position within the obstacle's influence range is calculated, and this direction is used as the guiding direction of the individual's obstacle avoidance velocity, guiding the individual away from the local extreme point.

[0126] The specific implementation of the obstacle avoidance and speed guidance method using the limit ring construction is as follows:

[0127] Establish a coordinate system with the center of the obstacle as the origin to describe the position and velocity information of the UAV. Obstacle coordinate system O xy With the geodetic coordinate system G xy The conversion relationship is as follows:

[0128] O xy =G xy -[x o ,y o ] T

[0129] In the formula, [x o ,y o ] T The coordinates of the obstacle in the geodetic coordinate system.

[0130] Based on the velocity direction of the UAV, a linear equation is constructed and simplified to the standard linear form, as follows:

[0131] ax + by + c = 0

[0132] The formula for calculating the closest distance d between line l and the center of the circle is as follows:

[0133]

[0134] Substituting d into the limit cycle equation, we obtain the direction of the guiding velocity.

[0135]

[0136]

[0137] In the formula, the asymptotic factor The degree of rotation is determined by d. The direction of obstacle avoidance is determined by d; when d is positive, the limit cycle rotates clockwise, and when d is negative, the limit cycle rotates counterclockwise. The radius of convergence of the limit cycle is r. s =r o +d o .

[0138] Transform the ideal velocity into a geodetic coordinate system, and rotate the velocity of UAV i to the direction of that ideal velocity to obtain the obstacle avoidance guidance velocity:

[0139]

[0140] Therefore, the input for obstacle avoidance control of individual clusters is:

[0141]

[0142] The drone swarm obstacle avoidance method of the present invention takes into account both obstacle avoidance and swarm formation requirements, and can safely pass through complex obstacle environments. Consider a swarm of 10 drones with initial positions of [2.9, 0.7; 16.7, -4.55; 7.5, -1.1; 10.4, 7.2; 8.3, -6.1; 6.6, 4.0; 3.6, -4.3; 11.3, 2.2; 12, -2.8; 15.9, 0.4] m and initial velocities of [3.427, -0.33; 3.27, -0.389; 3.48, -0.2; 3.28, -0.27; 3.58, 0.046; 3.49, -0.36; 3.32, -0.224; 3.589, -0.154; 3.46, -0.082; 3.533, -0.344] m / sv. max =5m / s, u max =10m / s. Set up simulation scenario 1, where the obstacle is located at the center of the line connecting the cluster and the target point. At this time, the velocity direction of the drones in the stable cluster form is perpendicular to the tangent plane of the obstacle. The cluster is very likely to get stuck in the local extreme value region, which is the most difficult obstacle avoidance situation in the cluster obstacle scenario. Figure 6 A schematic diagram of the obstacle avoidance trajectory of the cluster is provided. Simulation scenario 2 is set up: the obstacle is located to the side of the cluster's path, in which case local extrema are less likely to occur. Figure 7 Provide a schematic diagram of the overall obstacle avoidance trajectory of the cluster. Set up simulation scenario 3, and set multiple obstacles of different positions and sizes relative to the cluster in the original driving path. The obstacles are sparsely distributed to ensure that the threat range of each obstacle does not overlap. Figure 8 Provide the obstacle avoidance trajectory for the cluster. Figure 9 The curves showing the change in distance between the drones that eventually become neighbors are presented. Figures 10-11 Provide the speed variation curve of the UAV.

[0143] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for obstacle avoidance in a quadrotor UAV swarm based on vision-of-view and speed-guided approaches, characterized in that, Includes the following steps: Step 1: Based on the UAV motion model, design the UAV control input, which includes UAV swarm behavior motion control terms; Step 2: Design an adaptive communication topology mechanism for UAV swarms based on the field-of-view method; Step 3: Construct smooth adaptive communication weight coefficients based on the neighbor's field of view and relative distance; Step 4: Based on the "far repulsion, near attraction" principle, represent the motion control terms of the drone swarm behavior; Step 5: Design the obstacle avoidance guidance velocity term based on the limit cycle method, and construct the external obstacle avoidance control input for the UAV swarm using the artificial potential field method; The UAV motion model mentioned in step 1 is as follows: ； in, For drones The two-dimensional spatial position vector, with the superscript "·" representing the derivative. For drones The two-dimensional spatial velocity vector, For drones The control inputs are x and y, which are the x-axis and y-axis in two-dimensional space, respectively, and T is the transpose operation; The implementation process of step 2 is as follows: First, calculate the drone relative to drones Angle between the directions of motion : ； in Let be the two-dimensional spatial position vector of UAV j, and the symbol "·" represents the dot product; Next, determine the included angle. Does it belong to drones? Field of vision Inside: ； With drones Based on speed, it spreads to both sides. angle; Determine if a drone is in use With drones Is the distance between them less than the communication distance? : ； There is an adaptive communication topology mechanism for drone swarms. for: ； Step 3 is implemented by selecting a smoothing function. : ； In the formula, The smoothing coefficient is z; z is a parameter that determines the degree of smoothing. Constructing smooth adaptive communication weight coefficients : ； in, The field of view weighting coefficient indicates that the closer an individual is to the extended line of the body's velocity, the greater the influence of the body on the individual. This is a distance weighting coefficient, indicating that the closer an individual is to the host, the greater the influence of the host. The host and the individual refer to the drone, respectively. and drones .

2. The obstacle avoidance method for quadrotor UAV swarms based on vision-of-view and speed guidance as described in claim 1, characterized in that, The speed and control input of the UAV motion model satisfy the following: ； in," "" indicates taking the L2 norm, and the control input is based on the UAV motion model. Represented as: ； in For the motion control of drone swarm behavior, To avoid external obstacles.

3. The obstacle avoidance method for quadrotor UAV swarms based on vision-of-view and speed guidance according to claim 2, characterized in that, In step 4, the motion control terms for drone swarm behavior are divided into two parts: ； in, This serves as the input for coordinated collision avoidance speed control within a drone swarm. The attraction of the target point to the drone swarm and the ideal speed of the drone swarm; ； in, and All are adjustable parameters; Indicates drone Pointing to drones unit vector, (·) represents the power function: ； Where a, b, and c are all parameters; ； in, The target point location, For the ideal movement speed of the drone swarm, meet ; and These are all parameter adjustments; all drones in the drone swarm are aligned with the target point and the ideal speed. The motion control term for drone swarm behavior is then expressed as: 。 4. The obstacle avoidance method for quadrotor UAV swarms based on vision-of-view and speed guidance according to claim 3, characterized in that, The implementation process of step 5 is as follows: First, the obstacle is made convex, the minimum circumcircle of the obstacle is drawn, and the obstacle threat radius r is set. s The drone swarm enters the obstacle threat radius r s When inside, it will be subject to the repulsive force of obstacles. Based on the drone's position information, a repulsive force function is constructed. Find the repulsive force on the drone: ； In the formula, β refers to the coefficient related to obstacle avoidance. For drone i, the distance to the first The nearest location of each obstacle, with a weighting factor. The range of the repulsive force is smoothed. For the obstacle avoidance vector direction, Here is the expression for the repulsive force function; Establish a coordinate system with the center of the obstacle as the origin to describe the position and velocity information of the UAV. With the geodetic coordinate system The conversion relationship is as follows: ； In the formula, The coordinates of the obstacle in the geodetic coordinate system; Based on the velocity direction of the UAV, a linear equation is constructed and simplified to the standard linear form, as follows: Where a, b, and c are parameters; Calculate the line closest distance to the center of the circle The calculation formula is as follows: ； Will Substituting into the limit cycle equation, we obtain the direction of the guiding velocity. : ； In the formula, the asymptotic factor It determines the degree of rotation and distance. The direction of obstacle avoidance has been determined, when When the value is positive, the limit cycle rotates clockwise. When the value is negative, the limit cycle rotates counterclockwise; and the radius of convergence of the limit cycle is set to r. s ; Transform the ideal velocity v0 to the geodetic coordinate system, and then... (The sentence is incomplete and requires more context to translate accurately. It appears to be referring to a drone.) Speed ​​rotation v i By rotating to the ideal velocity v0 direction, we can obtain the guiding velocity of the i-th drone under the influence of the k-th obstacle. : ； The external obstacle avoidance control input for the drone swarm is: ； in, and These are all adjustable parameters.

Images