A UAV cooperative control method and system

Abstract

The present invention relates to the field of unmanned aerial vehicle flight control, and in particular to a method and system for cooperative control of unmanned aerial vehicles. The self-organizing network cooperative control method of the present invention can take into account unmanned aerial vehicles with different power attenuation conditions, match unmanned aerial vehicles with different maneuverability performances, and allocate a matching group of unmanned aerial vehicles with mutually matching maneuverability performances to topological points of similar paths according to the similarity of the paths, and allocate unmanned aerial vehicles with maneuverability performances that do not match those of other unmanned aerial vehicles to non-matching paths, so that the performance of unmanned aerial vehicles can be maximized under existing conditions, and the action consistency of a group of unmanned aerial vehicles can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicles, and in particular to a method and system for cooperative control of an unmanned aerial vehicle ad hoc network. Background Art

[0002] With the development of drones, the application market of drones has grown explosively. Drones are most commonly used in the field of remote sensing, including land use and land cover changes, forest management, hydrological resource monitoring, weather forecasting and disaster response.

[0003] With the booming consumer market, the drone performance market has achieved strong growth. The demand for the application of drone formation flight and networking control is becoming increasingly strong.

[0004] However, when drones are operated in swarms or small formations, the number of drones increases exponentially, and cost control requires that they cannot all use high-cost drones. While reducing costs, ensuring accurate control of the formation has become a difficult problem in the field of drone applications. However, due to factors such as quality control, the performance degradation of low-cost drones during flight varies greatly, especially the degradation of battery and drive performance.

[0005] In addition, during the flight of the UAV, it may face interference factors from the external environment or differences in the UAV's own conditions, which will affect the control effect of the formation flight. Summary of the invention

[0006] In view of the above problems existing in the prior art, the present invention proposes a highly efficient self-organizing network collaborative control method and system. The self-organizing network collaborative control method of the present invention can take into account drones with different power attenuation conditions, match drones with different maneuverability, and assign a matching group of drones with mutually matching maneuverability to topological points of similar paths according to the similarity of the paths, and assign drones that do not match the maneuverability of other drones to non-matching paths, so as to maximize the use of drone performance under existing conditions, and replace drones whose maneuverability cannot meet the matching conditions at any time, and assign them to non-matching paths or topological points with lower maneuverability requirements. In addition, the present invention can minimize the discreteness of flight deviations during flight.

[0007] Specifically, the present invention provides a method for cooperative control of unmanned aerial vehicles, the method comprising the following steps:

[0008] (1) Determine the mass of each drone in the drone formation, the target topological structure of the drone formation flight, and the flight trajectory of each topological point in the topological structure;

[0009] (2) Start each UAV. When each UAV rises to a predetermined height, issue a basic flight command to all UAVs, measure the basic maneuverability index of each UAV under the basic flight command, construct a basic weighted maneuverability index of each UAV based on the basic maneuverability index and the mass of each UAV, and express the basic weighted maneuverability index as a basic weighted maneuverability vector;

[0010] in,

[0011]

[0012] H v , S v ,φ v , θ v They represent the weighted maneuvers of vertical lift, horizontal flight, roll, yaw, and pitch, respectively; Δt1-Δt5 represent the time to complete the corresponding actions, φ2, φ1 represent the final and initial roll angles, respectively; θ2, θ1 represent the final and initial pitch angles, respectively. Respectively represent the final and initial yaw angles, m represents the mass of the UAV, P1-P5 represent the total output power of the UAV to complete the corresponding action, P0 represents the basic output power in the hovering state, δ1-δ5 represent the weighting coefficients of each weighted maneuverability;

[0013] (3) Based on the initial flight trajectory of each topological point, each flight trajectory is divided, and the topological structure stability period exceeding the predetermined maintenance time is used as a division node to divide the flight process into multiple flight time periods;

[0014] (4) For each flight period, the matching degree of the initial flight trajectory of each topological point in the period is calculated, and the flight trajectory matching degree is divided. The topological points with matching degrees higher than a predetermined value are divided into one or more matching groups, and the remaining topological points are divided into non-matching groups;

[0015] (5) For each UAV, determine the distance between the maneuverability vectors of each UAV, and based on the difference of the maneuverability vectors of the UAVs, select the corresponding number of UAVs with the smallest maneuverability vector difference from low to high for each matching group;

[0016] (6) During the flight of the drone, the output power, coordinate position and expected target position of each drone are obtained at predetermined time intervals or in real time, and then the real-time equivalent maneuverability vector of each drone is calculated. The difference between the real-time maneuverability vector of each drone and the average maneuverability vector in each matching group is calculated. When the difference between the real-time maneuverability vector and the average maneuverability vector of any drone exceeds a threshold, it is regarded as a downgraded drone, and a matching group switching operation is performed on the downgraded drone. The candidate drone with the smallest difference with the average maneuverability vector of the matching group is selected from the non-matching group and promoted to the matching group drone.

[0017] In a preferred implementation, step (6) includes: selecting a number of candidate drones from the non-matching group whose average mobility vector difference with the matching group is the smallest, and calculating the comprehensive path cost required for each candidate drone to switch to the matching group with the degraded drone;

[0018] The candidate UAV with the lowest comprehensive path cost is selected to switch the flight path with the downgraded UAV, and the flight paths of the two are replanned.

[0019] In another preferred implementation, the topological structure stable period is when the change amplitude of the position or position relationship of more than 60-90% of all topological points in a period of more than 0.5s-2s is less than the set value.

[0020] In another preferred implementation, the basic flight instructions include one or more of vertical ascent, yaw, pitch, horizontal advance, and roll.

[0021] In another preferred implementation, the real-time equivalent mobility vector is calculated based on the following formula: The real-time equivalent mobility vector is calculated based on the following method,

[0022] in,

[0023]

[0024] H' v , S' v 、φ' v , θ' v They represent the real-time weighted maneuverability of vertical lift, horizontal flight, roll, yaw, and pitch, respectively. Δt' represents the integration time step, and the time step of each parameter is the same. φ'2, φ'1 represent the roll angle at the end and beginning of the current calculation cycle, respectively. θ'2, θ'1 represent the pitch angle at the end and beginning of the current calculation cycle, respectively. They represent the yaw angles at the end and beginning of the current calculation cycle, m represents the mass of the drone, and P' sumThey represent the total output power for completing the current integral step action, and R1-R5 represent the ratio of the power consumed by the corresponding maneuver to the total power.

[0025] On the other hand, the present invention provides a UAV self-organizing network collaborative control system, the system comprising a processor and a memory, and when the processor executes a flight control program stored in the memory, the processor executes the described method.

[0026] Beneficial Effects

[0027] The self-organizing network collaborative control method of the present invention can take into account drones with different power attenuation conditions, match drones with different maneuverability, and assign a matching group of drones with maneuverability that matches each other to topological points of similar paths according to the similarity of the paths, and assign drones that do not match the maneuverability of other drones to non-matching paths. It can maximize the performance of drones under existing conditions, and replace drones whose maneuverability cannot meet the matching conditions at any time, and assign them to non-matching paths or topological points with lower maneuverability requirements. In addition, the present invention can maximize the consistency of drone actions. During the drone flight performance, the consistency of the actions of drones with similar paths will greatly affect the viewing experience, and the present invention uses matching groups to complete the flight of similar paths, which can greatly improve the consistency of the actions of the matching groups. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a schematic flow chart of the method of the present invention.

[0029] Figure 2 These are the test data of the control methods of group A and group B in the embodiments of the present invention in terms of average deviation and deviation dispersion. DETAILED DESCRIPTION

[0030] The present invention will be described in detail below in conjunction with the accompanying drawings and embodiments thereof, but the protection scope of the present invention is not limited to the scope of the embodiments described herein.

[0031] The collaborative control method in this embodiment includes the following steps:

[0032] (1) Determine the mass of each drone in the drone formation, the target topological structure of the drone formation flight, and the flight trajectory of each topological point in the topological structure;

[0033] (2) starting each UAV, and when each UAV rises to a predetermined height, issuing basic flight instructions to all UAVs, and measuring basic maneuverability indexes of lift, rotation, and pitch of each UAV under the basic flight instructions, wherein the basic flight instructions at least include the total output power of each rotor of each UAV, constructing a basic weighted maneuverability index of each UAV based on the basic maneuverability index and the mass of each UAV, and expressing the basic weighted maneuverability index as a basic weighted maneuverability vector;

[0034] in,

[0035]

[0036] H v , S v ,φ v , θ v They represent the weighted maneuvers of vertical lift, horizontal flight, roll, yaw, and pitch, respectively; Δt1-Δt5 represent the time to complete the corresponding actions, φ2, φ1 represent the final and initial roll angles, respectively; θ2, θ1 represent the final and initial pitch angles, respectively. They represent the final and initial yaw angles respectively, m represents the mass of the UAV, and P1-P5 represent the total output power of the UAV to complete the corresponding actions.

[0037] (3) Based on the initial flight trajectory of each topological point, each flight trajectory is divided, and the topological structure stability period exceeding the predetermined maintenance time is used as a division node to divide the flight process into multiple flight time periods;

[0038] (4) For each flight period, the matching degree of the initial flight trajectory of each topological point in the period is calculated, and the flight trajectory matching degree is divided. The topological points with matching degrees higher than a predetermined value are divided into one or more matching groups, and the remaining topological points are divided into non-matching groups.

[0039] In a preferred implementation, the flight trajectory matching degree is calculated using the Discrete Frechet Distance method, or it can also be calculated using the DTW method. Both methods are methods for calculating trajectory similarity, and the higher the trajectory similarity, the higher the trajectory matching degree.

[0040] (5) For each drone, determine the distance between the basic weighted maneuverability vectors of each drone. Based on the distance difference of the basic weighted maneuverability vectors of the drones, select the corresponding number of drones with the smallest maneuverability vector difference from low to high for each matching group. Take the simplest one matching group and one non-matching group as an example. Assume that the number of drones is 100. For one flight phase, assume that the number of matching groups with topological structure similarity higher than the threshold in this phase is 60, and the number of topological points in the non-matching group is 40. Calculate the basic weighted maneuverability vectors of the 100 drones. For the 100 drones, select 60 drones, calculate the average maneuverability vector of the 60 drones, calculate the distance between the maneuverability vector of any one of the 60 drones and the average maneuverability vector, select the drone with the largest distance from the average maneuverability vector, and replace it with the drone with a distance from the average maneuverability vector smaller than that of the remaining 40 drones. Recalculate the average maneuverability vector, and repeat this step until the average distance between the 60 drones and the average maneuverability vector is the smallest. The 60 drones are taken as the matching group, and the remaining drones are taken as the non-matching group.

[0041] Preferably, before this, for the topological points in the non-matching group, the maneuverability requirements of the corresponding UAV trajectories in the non-matching group are determined, and for flight trajectories with maneuverability vectors higher than a certain value, high-maneuverability UAVs are preferentially matched for them, and the remaining UAVs are used to match the trajectories in the matching group.

[0042] In a preferred implementation, the maneuverability requirement of the trajectory is weighted by calculating the turning radius requirement, the climbing rate requirement, etc. In another preferred implementation, the maneuverability requirement of the trajectory is determined based on an index corresponding to the above-mentioned measurement index.

[0043] (6) During the flight of the drone, the output power, coordinate position and expected target position of each drone are obtained at predetermined time intervals or in real time, and then the real-time equivalent maneuverability vector of each drone is calculated. The distance between the real-time maneuverability vector of each drone and the average maneuverability vector in each matching group is calculated. When the distance between the real-time maneuverability vector of any drone and the average maneuverability vector exceeds a threshold, it is regarded as a downgraded drone, and a matching group switching operation is performed on the downgraded drone. Several candidate drones with the smallest difference in average maneuverability vector with the matching group are selected from the non-matching group, and the comprehensive path cost required for each candidate drone to switch with the downgraded drone is calculated.

[0044] In a preferred implementation, the threshold is set to 2-3 times the distance between the real-time maneuverability vector of each UAV in the matching group and its average maneuverability vector.

[0045] In a preferred implementation, the real-time equivalent mobility vector is calculated based on the following method:

[0046] in,

[0047]

[0048] H' v , S' v 、φ' v , θ' v They represent the real-time weighted maneuverability of vertical lift, horizontal flight, roll, yaw, and pitch, respectively. Δt' represents the integration time step, where the time step of each parameter is the same. φ'2, φ'1 represent the roll angle at the end and beginning of the current calculation cycle, respectively. θ'2, θ'1 represent the pitch angle at the end and beginning of the current calculation cycle, respectively. They represent the yaw angles at the end and beginning of the current calculation cycle, m represents the mass of the drone, and P' sum They represent the total output power of the current integral step action, R1-R5 represent the proportion of power consumption of the corresponding maneuvering action, which can be estimated based on the difference between the power consumed by completing the specified action per unit distance and unit angle and the power consumed in the static hovering state under a single action, divided by the ratio of the power consumed in the static hovering state for the same time, and P0 represents the power consumption of the corresponding hovering time. Taking R1 as an example, In addition, it should be noted that the basic weighted maneuverability index and the real-time weighted maneuverability index in the present invention are relative quantities, and their main purpose is to determine the changes and similarities in maneuverability of each UAV. Preferably, they can be normalized.

[0049] (7) Select the candidate UAV with the lowest comprehensive path cost and the downgraded UAV to switch the flight path, and replan the flight paths of the two.

[0050] Preferably, the method further comprises: the topological structure stable period is when the change amplitude of the position or position relationship of more than 60-90% of all the topological points within a period of more than 0.5s-2s is less than a set value.

[0051] Preferably, the basic flight instructions include at least one or more of vertical ascent, yaw, pitch, horizontal advance, and roll.

[0052] Preferably, the basic maneuverability index at least includes the ascent and descent rate Ev, the rotation rate Rv of the fuselage, the turning efficiency Tv, and the pitch adjustment speed Pv, the maneuverability vector includes a vector composed of multiple of the above indexes, and the maneuverability vector difference is determined based on the Euclidean distance between each vector.

[0053] Preferably, the actual maneuverability of the UAV is determined based on the deviation between the actual completion status and the expected completion status.

[0054] In a preferred implementation, a PSO algorithm (Particle Swarm Optimization) is used for initial path planning, and in step (7), the PSO algorithm is used to re-plan the paths of the candidate UAVs and the downgraded UAVs.

[0055] Specifically, each drone is assumed to be a particle with a certain mass. The specific path in the drone's flight requirements is used as a constraint, and the same number of "particles" (representing candidate solutions) as the number of drones are used to search for the global optimal solution in space. Each particle is set with at least two key attributes: position and velocity. In each iteration, each particle updates its velocity and position based on its own historical best position (pBest) and the historical best position of the entire group (gBest), so as to gradually approach a better solution.

[0056] The speed and position update process of each particle can be expressed as follows:

[0057] Vi(t+1)=ωVi(t)+c1γ1(pBest i (t)-X i (t))+c2γ2(gBesti(t)-Xi(t))

[0058] Xi(t+1)=Xi(t)+Vi(t+1):

[0059] Where Vi(t) is the velocity vector, Xi(t) is the position vector, which respectively represent the velocity and position of particle i at time t, Vi(t+1) and Xi(t+1) respectively represent the velocity and position of particle i at time t+1. Ω is the inertia factor, c 1、 c2 are acceleration factors, pBest i (t) and gBesti(t) represent the historical best position and the global best position.

[0060] In a preferred implementation, during path planning, the necessary path (specific path) and the connection path required by the flight mission are planned separately. For example, a specific path requires all drones to form a ring and fly around the center of the circle three times. The part flying around the ring is called the specific path, and the path from the previous topological shape to the ring is called the connection path. For the specific path, since it is the task path specified by the flight mission, the path cannot be adjusted. It is only necessary to use the conventional path planning algorithm to plan according to the trajectory requirements of the time set by the task. When planning the connection path, it is determined whether there is a downgrade and upgrade operation of the matching drone. If so, when planning the connection path, for each downgraded drone, if there are multiple candidate substitute drones, the path planning is performed separately in the manner that each downgraded drone corresponds to a candidate substitute drone. Assuming there are 3 candidate drones, candidate drone 1 is used to replace the downgraded drone for a path planning, candidate drone 2 is used to replace the downgraded drone for a path planning, and candidate drone 3 is used to replace the downgraded drone for a path planning. After completing three plannings, the comprehensive cost of the three path plannings is calculated.

[0061] Comprehensive cost C T The calculation can be done in the following way:

[0062] C T =η1×C d +η2×C e +η3×C H +η4×C t

[0063] C d represents the comprehensive distance cost, which is obtained by weighted summing of the switching lengths of all UAV paths;

[0064] C e It represents the energy consumption cost, which is obtained by estimating the energy consumption of all UAV paths based on the energy consumption of each UAV in the benchmark flight state;

[0065] C H represents the flight altitude change cost, which is obtained by summing the flight altitude changes of all UAVs;

[0066] C t It represents the time cost, which is obtained by calculating the variance between the time required for all drones to complete the switching action and the average time required. The smaller the variance, the lower the cost.

[0067] Based on the comprehensive costs of different switching schemes, the switching scheme with the lowest cost is selected.

[0068] In order to verify the effect of the self-organizing network collaborative control system of the present invention in collaborative control, 10 small quadrotor drones with different usage time were used, of which 6 adopted similar trajectories for circular flight, and the remaining 4 adopted random paths for the same path length but different trajectory shapes (that is, the matching group contained 6 drones and the non-matching group contained 4 drones). The control effects of the drones in two states, namely the control mode based on the matching group (A) and the control mode based on the non-matching group (B), were measured respectively. The control deviation and deviation discreteness parameters under the two control modes of groups A and B were measured. The flight control was implemented based on the PX4 open source flight control program. Group A added a matching group plug-in, and group B did not add a plug-in.

[0069] The control deviation is calculated using the following formula:

[0070] k is the number of drones, n is the number of samples in the control path, i∈[1,n], (x i ',y i ',z i ') is the target position of the UAV at the i-th sampling time in the initial path planning, (x i ,y i ,z i ) is the actual position of the UAV at the i-th sampling time. The UAV is kept flying continuously and sampling is performed every 3 minutes to check the control of the UAV deviation during the continuous flight.

[0071] For any sampling time, (x i '-x i ,y i '-y i ,z i '-z i ) constructs the deviation vector and calculates the average variance between each deviation vector and the mean of the deviation vector. The average variance can reflect the discreteness of the position deviation of the UAV during flight. The smaller the variance, the smaller the discreteness. Since there is no absolute reference object during the flight of the UAV, the smaller the discreteness of the flight deviation of the UAV is, the smaller the visual impact is.

[0072] For the drone in the above embodiment, 6 sampling time points are taken to measure the average deviation of group A and group B and the deviation dispersion of each sampling point.

[0073] The results of the test are as follows Figure 2 As shown. Figure 2 It can be seen from the comparative data in that after adding the control method of the present invention as a control aid, although the improvement of the average deviation is limited, the deviation dispersion at each sampling point is significantly reduced.

[0074] Although the principles of the present invention are described in detail above in conjunction with the preferred embodiments of the present invention, those skilled in the art should understand that the above embodiments are merely explanations of the exemplary implementations of the present invention, and are not intended to limit the scope of the present invention. The details in the embodiments do not constitute limitations on the scope of the present invention, and any obvious changes such as equivalent transformations, simple replacements, etc. based on the technical solution of the present invention, without departing from the spirit and scope of the present invention, fall within the protection scope of the present invention.

Claims

1. A method for cooperative control of unmanned aerial vehicles, characterized in that the method comprises the following steps: (1) Determine the mass of each drone in the drone formation, the target topological structure of the drone formation flight, and the flight trajectory of each topological point in the topological structure; (2) Start each UAV. When each UAV rises to a predetermined height, issue basic flight instructions to all UAVs and measure the basic maneuverability index of each UAV under the basic flight instruction. Based on the basic maneuverability index and the mass of each UAV, construct the basic weighted maneuverability index of each UAV. Express the basic weighted maneuverability index as a basic weighted maneuverability vector in, H v , S v ,φ v , θ v They represent the weighted maneuvers of vertical lift, horizontal flight, roll, yaw, and pitch, respectively; Δt1-Δt5 represent the time to complete the corresponding actions, φ2, φ1 represent the final and initial roll angles, respectively; θ2, θ1 represent the final and initial pitch angles, respectively. Respectively represent the yaw angles at the end and the beginning, m represents the mass of the UAV, P1-P5 represent the total output power of the UAV to complete the corresponding action, P0 represents the basic output power in the hovering state, δ1-δ5 represent the weighting coefficients of each weighted maneuverability, x1, y1, z1 are the actual positions of the UAV at the first sampling time, and x2, y2, z2 are the actual positions of the UAV at the second sampling time; (3) Based on the initial flight trajectory of each topological point, each flight trajectory is divided, and the topological structure stability period exceeding the predetermined maintenance time is used as the division node, and the flight process is divided into multiple flight time periods; (4) For each flight period, the matching degree of the initial flight trajectory of each topological point in the period is calculated, and the flight trajectory matching degree is divided. The topological points with matching degrees higher than a predetermined value are divided into one or more matching groups, and the remaining topological points are divided into non-matching groups; (5) For each UAV, determine the distance between the mobility vectors of each UAV, and based on the difference of the UAV mobility vectors, select the corresponding number of UAVs with the smallest mobility vector difference from low to high for each matching group; (6) During the flight of the UAV, the output power, coordinate position and expected target position of each UAV are obtained at predetermined time intervals or in real time, and then the real-time equivalent maneuverability vector of each UAV is calculated. In each matching group, the distance between the real-time maneuverability vector of each UAV and the average maneuverability vector in the matching group and the average distance between the real-time maneuverability vector of each UAV and the average maneuverability vector in the matching group are calculated. When the distance between the real-time maneuverability vector of any UAV and the average maneuverability vector exceeds a predetermined multiple of the average distance, it is regarded as a downgraded UAV, and a matching group switching operation is performed on the downgraded UAV. The candidate UAV with the smallest distance to the average maneuverability vector of the matching group is selected from the non-matching group and promoted to the matching group UAV.

2. The UAV cooperative control method according to claim 1, characterized in that: The step (6) comprises: selecting a number of candidate drones from the non-matching group whose distances to the average mobility vector of the matching group are the smallest, and calculating the comprehensive path cost required for each candidate drone to switch to the matching group with the degraded drone; The candidate UAV with the lowest comprehensive path cost is selected to switch the flight path with the downgraded UAV, and the flight paths of the two are replanned.

3. The UAV cooperative control method according to claim 1, characterized in that: The topological structure stable period is when the change range of the position or position relationship of more than 60-90% of all topological points in a period of more than 0.5s-2s is less than the set value.

4. The UAV cooperative control method according to claim 1, characterized in that: The basic flight instructions include one or more of vertical ascent, yaw, pitch, horizontal advance, and roll.

5. The UAV cooperative control method according to claim 1, characterized in that: The real-time equivalent mobility vector is calculated based on the following formula: The real-time equivalent mobility vector is calculated based on the following method, in, H' v , S' v 、φ' v , θ' v They represent the real-time weighted maneuverability of vertical lift, horizontal flight, roll, yaw, and pitch, respectively. Δt' represents the integration time step, and the time step of each parameter is the same. φ'2, φ'1 represent the roll angle at the end and beginning of the current calculation cycle, respectively. θ'2, θ'1 represent the pitch angle at the end and beginning of the current calculation cycle, respectively. They represent the yaw angles at the end and beginning of the current calculation cycle, m represents the mass of the drone, and P' sum They represent the total output power required to complete the current integral step action, R1-R5 represent the ratio of the power consumed by the corresponding maneuvering action to the total power, x1', y1', z1' are the target positions of the UAV at the first sampling time in the initial path planning, and x2', y2', z2' are the target positions of the UAV at the second sampling time in the initial path planning.

6. A UAV self-organizing network collaborative control system, characterized in that: The system comprises a processor and a memory. When the processor executes a flight control program stored in the memory, the processor executes the method according to any one of claims 1 to 5.

Images