Multi-unmanned aerial vehicle cooperative formation obstacle avoidance method based on improved artificial potential field

Abstract

The invention relates to the field of formation flight control, in particular to a multi-unmanned aerial vehicle cooperative formation obstacle avoidance method based on an improved artificial potential field. Firstly, mutual collision between the unmanned aerial vehicles in the formation is avoided through newly adding anti-collision repulsive force between the unmanned aerial vehicles; secondly, providing a local target point mechanism to prevent the piloting unmanned aerial vehicle from falling into a local minimum trap and further prevent the whole formation from falling into the local minimum trap; then, a dynamic repulsive force fusion strategy is introduced, oscillation of the unmanned aerial vehicle near the obstacle is reduced, and therefore the stability of the whole formation is improved; finally, a hiking optimization algorithm gravity coefficient adaptive controller is designed, and the formation pattern maintaining and recovering precision is improved. According to the invention, formation and obstacle avoidance are optimized, so that rapid formation and obstacle avoidance of multiple unmanned aerial vehicles in a complex environment are realized.

Description

Technical Field

[0001] This invention belongs to the field of formation flight control, specifically relating to a multi-UAV cooperative formation obstacle avoidance method based on an improved artificial potential field. Background Technology

[0002] Multi-UAV cooperative formation obstacle avoidance is a crucial research topic in formation flight control, widely applied in environmental monitoring, maritime search and rescue, and topographic mapping. In actual flight, UAV formations face challenges from complex obstacle environments, requiring effective obstacle avoidance. Maintaining and restoring formation after successful obstacle avoidance is equally critical; otherwise, effective communication is difficult to maintain, impacting mission efficiency. Existing multi-UAV formation obstacle avoidance systems still suffer from these issues.

[0003] To address the issues of insufficient accuracy in obstacle avoidance in complex environments and in maintaining and restoring formation, this invention provides a multi-UAV cooperative formation obstacle avoidance method based on an improved artificial potential field. This method improves the accuracy of UAV formation maintenance and restoration and enables effective formation obstacle avoidance in complex environments. Summary of the Invention

[0004] (1) Construct a three-dimensional motion model and formation model of the UAV;

[0005] (2) Construct the gravitational potential field model and repulsive potential field model of the UAV;

[0006] (3) Construct an inter-machine anti-collision repulsive potential field model to prevent collisions between UAVs;

[0007] (4) Establish a local target point mechanism to prevent UAV formations from falling into local minima traps;

[0008] (5) Design a dynamic repulsion fusion strategy to reduce the oscillation of the UAV and stabilize its motion process;

[0009] (6) Based on the hiking optimization algorithm, design an adaptive controller for the gravity coefficient of the hiking optimization algorithm;

[0010] (7) Construct a drone formation collaborative controller.

[0011] Specifically, step (1) involves constructing a three-dimensional motion model and a formation model for the UAV, and the specific steps are as follows:

[0012] (1.1) Based on the second-order integrator dynamics model, a multi-UAV formation model is established in three-dimensional space, where the second-order integrator is the first UAV formation model. The kinematic model of the drone is as follows:

[0013]

[0014] In the formula , , representing the system position vector, , representing the system velocity vector, This indicates the control input for the drone.

[0015] (1.2) Taking three UAVs as an example, an isosceles right triangle formation model was constructed. The relative positions of each UAV are as follows: Figure 2 As shown. Select The drone serves as the lead drone, while the others act as follower drones. Based on the geometric constraints and shape of the formation, the desired positions of each follower drone in the formation are calculated. Expected position As shown below.

[0016]

[0017] in, To follow the drone The expected distance relative to the pilot drone.

[0018] Specifically, the steps for constructing the gravitational and repulsive potential field model in step 2 are as follows:

[0019] (2.1) The positions of the UAV, the target point, and the obstacle are respectively defined as follows: , and The gravitational potential field function and the repulsive potential field function are as follows:

[0020]

[0021]

[0022] in, For gravitational field coefficients, The repulsive field coefficient is... The maximum range of the repulsive force field influence of the obstacle. This represents the distance from the current path point to the target point. Indicates the path from the current path point to the th The distance to the obstacle.

[0023] (2.2) By taking the negative gradient of the gravitational potential field function, we can obtain the gravitational force exerted by the target point on the current path point. As shown below:

[0024]

[0025] Taking the negative gradient of the repulsive potential field function yields the repulsive force exerted by the obstacle point on the current path point. As shown below:

[0026]

[0027] (2.3) The forces acting on each path point are the target gravitational force and the gravitational force. The resultant force of the repulsive forces generated by the obstacles is shown below:

[0028]

[0029] Specifically, the steps for constructing the inter-machine anti-collision repulsive potential field model in step 3 are as follows:

[0030] (3.1) When the drone With drones The distance between them is less than the set safety threshold. At that time, drones Subject to drones The resulting repulsive force, and the anti-collision repulsive force field function between the two, can be expressed as:

[0031]

[0032] In the formula , is the inter-machine anti-collision repulsion coefficient. This is the formation safety threshold.

[0033] (3.2) Taking the negative gradient of the above equation, we can obtain the inter-machine anti-collision repulsion force as:

[0034]

[0035] Specifically, step (4) of establishing the local target point mechanism is as follows:

[0036] (4.1) Define the local optimum problem as:

[0037]

[0038] Where, This is a very small proportional factor. The above formula means that when the resultant force of the target point's attraction, the obstacle's repulsion, and the inter-drone collision avoidance force is less than a certain minimum value, the navigation drone will fall into a local optimum.

[0039] (4.2) The local target point mechanism is as follows:

[0040]

[0041]

[0042] like Figure 3 As shown, it generates within a certain range of the pilot drone. The spherical surface contains a total of 441 local target points. For local target points To the obstacle distance, For local target points The overall cost, among which and The target point weight coefficient and obstacle weight coefficient are respectively, and both are constants. After calculating the comprehensive cost of all local target points, the one with the smallest comprehensive cost is selected as the next target point for the navigation drone.

[0043] Specifically, step (5) of establishing the dynamic repulsion fusion strategy is as follows:

[0044] (5.1) First, perform initialization:

[0045]

[0046]

[0047]

[0048] in The weighting factor coefficient is a constant. The net and repulsive forces of obstacles The initial weighting factor of the axis components, The coefficients are updated iteratively and are also constants. The net and repulsive forces of obstacles The maximum number of iterations for the axis component. for The repulsive force of obstacles at all times Axial components, for The repulsive force of obstacles at all times Initial iterative fusion of axis components.

[0049] (5.2) Then perform iterative updates.

[0050]

[0051]

[0052] in ; The net and repulsive forces of obstacles The first axis component Sub-iteration weight factor for The repulsive force of obstacles at all times The first axis component Subsequent iterations of fusion. After completing all iterations, the net and repulsive forces of the obstacles can be obtained. Final dynamic fusion of axis components Similarly, we can obtain Dynamic fusion of biaxial repulsive force components.

[0053] (5.3) The sum of the components yields Dynamic fusion repulsion at all times .

[0054]

[0055] Specifically, step (6), which involves designing the adaptive controller for the gravity coefficient of the hiking optimization algorithm, is as follows:

[0056] (6.1) The mathematical basis of the hiking optimization algorithm is the Tobler hiking function, as shown below:

[0057]

[0058] In the formula For the number of iterations Hikers speed, It refers to the slope of the terrain. As shown in the formula below:

[0059]

[0060] In the formula and These represent altitude and distance traveled by the hiker, respectively. This refers to the slope of the terrain. Then, the hiker's current speed is calculated.

[0061]

[0062] In the formula yes The uniformly distributed number; and They represent hikers The current velocity and the initial velocity. It's the position of team leader. hikers The scan factor is located at Within the range.

[0063] Finally, the hikers Update location It is given by the following formula:

[0064]

[0065] Hiker's location The initialization depends on the upper bound of the solution. and the lower realm Expressed by an equation:

[0066]

[0067] In the formula yes The uniformly distributed number within.

[0068] (6.2) To address the control accuracy issue of UAV formations, the gravity coefficient is selected as the parameter to be optimized, and the fitness function is constructed as follows:

[0069]

[0070] In the formula To follow the desired position of the drone, we first use a second-order integral model and the current control input of the lead drone to obtain the position of the lead drone at the next moment. Then, we combine geometric constraints and formation shape to obtain the desired position of the follow drone at the next moment. To predict the position of the drone in the next moment, we combine the second-order integral model with the control input of the drone at this moment to obtain the predicted position of the drone under a specific gravity coefficient.

[0071] Specifically, step (7) of constructing the UAV formation cooperative controller is as follows:

[0072] Based on the above improvements, the control input for the UAV is calculated as follows:

[0073]

[0074] In the formula As the positive control gain used for damping, the improved artificial potential field method has the following effect on UAV formation: Figure 4 As shown.

[0075] The above improvements ensure that multiple UAVs can effectively form up and avoid obstacles in complex environments while maintaining high accuracy in formation maintenance and recovery. Attached Figure Description

[0076] Appendix Figure 1 This is a flowchart of the present invention;

[0077] Appendix Figure 2 This is a formation information diagram in the present invention;

[0078] Appendix Figure 3 This is a schematic diagram of the local target point mechanism in this invention;

[0079] Appendix Figure 4 This is a schematic diagram of the improved artificial potential field method in this invention;

[0080] Appendix Figure 5 This is the cooperative formation obstacle avoidance diagram in this invention; Detailed Implementation

[0081] The following description, in conjunction with the accompanying drawings, further illustrates a multi-UAV cooperative formation obstacle avoidance method based on an improved artificial potential field provided by the present invention.

[0082] This invention provides a multi-UAV cooperative formation obstacle avoidance method based on an improved artificial potential field, comprising the following steps:

[0083] Step (1) Construct a 3D motion model and formation model of the UAV;

[0084] (1.1) Based on the second-order integrator dynamics model, a multi-UAV formation model is established in three-dimensional space, where the second-order integrator is the first UAV formation model. The kinematic model of the drone is as follows:

[0085]

[0086] In the formula , , representing the system position vector, , representing the system velocity vector, This indicates the control input for the drone.

[0087] (1.2) Taking three UAVs as an example, an isosceles right triangle formation model was constructed. The relative positions of each UAV are as follows: Figure 2 As shown. Select The drone serves as the lead drone, while the others act as follower drones. Based on the geometric constraints and shape of the formation, the desired positions of each follower drone in the formation are calculated. Expected position As shown below.

[0088]

[0089] in, To follow the drone The expected distance relative to the pilot drone.

[0090] Step (2) Construct the gravitational potential field model and repulsive potential field model of the UAV;

[0091] (2.1) The positions of the UAV, the target point, and the obstacle are respectively defined as follows: , and The gravitational potential field function and the repulsive potential field function are as follows:

[0092]

[0093]

[0094] in, For gravitational field coefficients, The repulsive field coefficient is... The maximum range of the repulsive force field influence of the obstacle. This represents the distance from the current path point to the target point. Indicates the path from the current path point to the th The distance to the obstacle.

[0095] (2.2) By taking the negative gradient of the gravitational potential field function, we can obtain the gravitational force exerted by the target point on the current path point. As shown below:

[0096]

[0097] Taking the negative gradient of the repulsive potential field function yields the repulsive force exerted by the obstacle point on the current path point. As shown below:

[0098]

[0099] (2.3) The forces acting on each path point are the target gravitational force and the gravitational force. The resultant force of the repulsive forces generated by the obstacles is shown below:

[0100]

[0101] Step (3) Construct an inter-machine anti-collision repulsive potential field model to prevent collisions between UAVs;

[0102] (3.1) When the drone With drones The distance between them is less than the set safety threshold. At that time, drones Subject to drones The resulting repulsive force, and the anti-collision repulsive force field function between the two, can be expressed as:

[0103]

[0104] In the formula , is the inter-machine anti-collision repulsion coefficient. This is the formation safety threshold.

[0105] (3.2) Taking the negative gradient of the above equation, we can obtain the inter-machine anti-collision repulsion force as:

[0106]

[0107] Step (4) Establish a local target point mechanism to prevent the UAV formation from falling into the local minimum trap;

[0108] (4.1) Define the local optimum problem as:

[0109]

[0110] Where, This is a very small proportional factor. The above formula means that when the resultant force of the target point's attraction, the obstacle's repulsion, and the inter-drone collision avoidance force is less than a certain minimum value, the navigation drone will fall into a local optimum.

[0111] (4.2) The local target point mechanism is as follows:

[0112]

[0113]

[0114] like Figure 3 As shown, it generates within a certain range of the pilot drone. The spherical surface contains a total of 441 local target points. For local target points To the obstacle distance, For local target points The overall cost, among which and The target point weight coefficient and obstacle weight coefficient are respectively, and both are constants. After calculating the comprehensive cost of all local target points, the one with the smallest comprehensive cost is selected as the next target point for the navigation drone.

[0115] Step (5) Design a dynamic repulsion fusion strategy to reduce the oscillation of the UAV and stabilize its motion.

[0116] (5.1) First, perform initialization:

[0117]

[0118]

[0119]

[0120] in The weighting factor coefficient is a constant. The net and repulsive forces of obstacles The initial weighting factor of the axis components, The coefficients are updated iteratively and are also constants. The net and repulsive forces of obstacles The maximum number of iterations for the axis component. for The repulsive force of obstacles at all times Axial components, for The repulsive force of obstacles at all times Initial iterative fusion of axis components.

[0121] (5.2) Then perform iterative updates.

[0122]

[0123]

[0124] in ; The net and repulsive forces of obstacles The first axis component Sub-iteration weight factor for The repulsive force of obstacles at all times The first axis component Subsequent iterations of fusion. After completing all iterations, the net and repulsive forces of the obstacles can be obtained. Final dynamic fusion of axis components Similarly, we can obtain Dynamic fusion of biaxial repulsive force components.

[0125] (5.3) The sum of the components yields Dynamic fusion repulsion at all times .

[0126]

[0127] Step (6) Based on the hiking optimization algorithm, design an adaptive controller for the gravity coefficient of the hiking optimization algorithm;

[0128] (6.1) The mathematical basis of the hiking optimization algorithm is the Tobler hiking function, as shown below:

[0129]

[0130] In the formula For the number of iterations Hikers speed, It refers to the slope of the terrain. As shown in the formula below:

[0131]

[0132] In the formula and These represent altitude and distance traveled by the hiker, respectively. This refers to the slope of the terrain. Then, the hiker's current speed is calculated.

[0133]

[0134] In the formula yes The uniformly distributed number; and They represent hikers The current velocity and the initial velocity. It's the position of team leader. hikers The scan factor is located at Within the range.

[0135] Finally, the hikers Update location It is given by the following formula:

[0136]

[0137] Hiker's location The initialization depends on the upper bound of the solution. and the lower realm Expressed by an equation:

[0138]

[0139] In the formula yes The uniformly distributed number within.

[0140] (6.2) To address the control accuracy issue of UAV formations, the gravity coefficient is selected as the parameter to be optimized, and the fitness function is constructed as follows:

[0141]

[0142] In the formula To follow the desired position of the drone, we first use a second-order integral model and the current control input of the lead drone to obtain the position of the lead drone at the next moment. Then, we combine geometric constraints and formation shape to obtain the desired position of the follow drone at the next moment. To predict the position of the drone in the next moment, we combine the second-order integral model with the control input of the drone at this moment to obtain the predicted position of the drone under a specific gravity coefficient.

[0143] Step (7) Construct a drone formation cooperative controller.

[0144] Based on the above improvements, the control input for the UAV is calculated as follows:

[0145]

[0146] In the formula As the positive control gain used for damping, the improved artificial potential field method has the following effect on UAV formation: Figure 4 As shown.

[0147] The above improvements ensure that multiple UAVs can effectively form up and avoid obstacles in complex environments while maintaining high accuracy in formation maintenance and recovery.

Claims

1. A multi-UAV cooperative formation obstacle avoidance method based on an improved artificial potential field, characterized in that, The steps include: Step 1: Construct a 3D motion model and formation model for the UAVs; Step 2: Construct the gravitational potential field model and repulsive potential field model of the UAV; Step 3: Construct an inter-drone anti-collision repulsive potential field model to prevent drones from colliding with each other; Step 4: Establish a local target point mechanism to prevent drone formations from getting trapped in local minima. Step 5: Design a dynamic repulsion fusion strategy to reduce the oscillation of the UAV and stabilize its motion. Step 6: Based on the hiking optimization algorithm, design an adaptive controller for the gravity coefficient of the hiking optimization algorithm; Step 7: Build a drone formation collaborative controller.

2. The multi-UAV cooperative formation obstacle avoidance method based on an improved artificial potential field according to claim 1, characterized in that... In step 1, the flight requirements of the UAV are considered, and a UAV kinematic model and a formation model are constructed.

3. The multi-UAV cooperative formation obstacle avoidance method based on an improved artificial potential field according to claim 1, characterized in that... In step 2, the requirements for drone obstacle avoidance and reaching the target point are considered, and a corresponding gravitational and repulsive potential field model is constructed.

4. The multi-UAV cooperative formation obstacle avoidance method based on an improved artificial potential field according to claim 1, characterized in that... In step 3, an inter-machine anti-collision repulsive potential field model was constructed based on the inter-machine anti-collision requirements.

5. The multi-UAV cooperative formation obstacle avoidance method based on an improved artificial potential field according to claim 1, characterized in that... In step 4, to prevent the formation from getting trapped in local optima, a local target point mechanism is established, as shown below: The above formula means that when the combined force of the target point attraction, obstacle repulsion, and inter-drone collision avoidance force is less than a certain minimum value, the navigation drone will fall into a local optimum. The local target point mechanism is as follows: For local target points To the obstacle distance, For local target points The overall cost, among which and The target point weight coefficient and obstacle weight coefficient are respectively, and both are constants. After calculating the comprehensive cost of all local target points, the one with the smallest comprehensive cost is selected as the next target point for the navigation drone.

6. The multi-UAV cooperative formation obstacle avoidance method based on an improved artificial potential field according to claim 1, characterized in that... In step 5, a dynamic repulsive force fusion strategy is established to stabilize the motion process, as shown below: First, initialize: in The weighting factor coefficient is a constant. The net and repulsive forces of obstacles The initial weighting factor of the axis components, The coefficients are updated iteratively and are also constants. The net and repulsive forces of obstacles The maximum number of iterations for the axis component. for The repulsive force of obstacles at all times Axial components, for The repulsive force of obstacles at all times Initial iterative fusion of axis components. Then iterative updates are performed. in ; The net and repulsive forces of obstacles The first axis component Sub-iteration weight factor for The repulsive force of obstacles at all times The first axis component Subsequent iterations of fusion. After completing all iterations, the net and repulsive forces of the obstacles can be obtained. Final dynamic fusion of axis components Similarly, we can obtain Dynamic fusion of biaxial repulsive force components. Adding the components together yields Dynamic fusion repulsion at all times .

7. The multi-UAV cooperative formation obstacle avoidance method based on an improved artificial potential field according to claim 1, characterized in that... In step 6, to improve formation maintenance and recovery accuracy, a trekking optimization algorithm gravity coefficient adaptive controller is designed.

8. The multi-UAV cooperative formation obstacle avoidance method based on an improved artificial potential field according to claim 1, characterized in that... In step 7, a drone formation cooperative controller is constructed to calculate the drone control input.

Images