Large-load unmanned aerial vehicle cooperative hoisting method, device, equipment and medium

Abstract

The invention provides a heavy-load unmanned aerial vehicle cooperative hoisting method, device and equipment and a medium, and relates to the technical field of edge calculation, and the method comprises the steps that a virtual pilot unmanned aerial vehicle in an unmanned aerial vehicle cluster calculates total load data of loads hoisted by a plurality of edge nodes based on load distribution data of each edge node; based on the total load data and the node position, determining an overall motion track of the unmanned aerial vehicle cluster, and sending the overall motion track to the edge node, so that the edge node finds a reference tension value needing to be provided and a node motion track based on an ant colony algorithm; if it is determined that the lifting rope of the first unmanned aerial vehicle in the multiple unmanned aerial vehicles is broken in the flight process, according to a preset load redistribution scheme, synchronous tension is distributed to each second unmanned aerial vehicle; and setting the first unmanned aerial vehicle as a proxy center node, and receiving an updated movement track of the proxy center node for the unmanned aerial vehicle cluster. According to the invention, the sudden fault in the cooperative lifting process of the multiple unmanned aerial vehicles can be quickly processed.

Description

Technical Field

[0001] The present application relates to the technical field of edge computing, and specifically to a method, device, equipment and medium for collaborative lifting of large-load drones. Background Art

[0002] Large-capacity UAVs are also called large UAVs. The Interim Regulations on the Flight Management of Unmanned Aerial Vehicles classify UAVs into five categories: micro, light, small, medium, and large, according to performance indicators such as weight. Among them, UAVs with a take-off mass greater than 150kg are classified as large UAVs.

[0003] Collaborative lifting of multiple large-load UAVs is an aerial transportation technology based on cluster control and load distribution of multi-rotor unmanned aerial vehicles. Its core lies in the fact that multiple UAVs, through flexible cables or rigid connection structures, jointly bear a large load or a load that exceeds the carrying capacity of a single aircraft, and complete the collaborative tasks of complex path planning, precise attitude control and dynamic load balance under the premise of ensuring flight stability. Its implementation depends on high-precision position perception systems, multi-source information fusion algorithms, and distributed or centralized control architectures. By introducing graph theory modeling, collaborative control laws, and robust control strategies, it can effectively deal with uncertain factors such as load swing, wind disturbance, and communication delay, further improve the stability and safety of the system, and expand its application prospects in emergency rescue, large-scale component assembly, and urban air logistics.

[0004] In the existing technology, the fault tolerance of drone lifting systems is weak in the face of sudden failures, especially non-structural failures such as rope breakage. This is mainly reflected in the system's lack of real-time perception and dynamic response mechanism for single-point suspension failures. Once a rope breaks, the remaining drones are often unable to quickly reconstruct the overall tension distribution strategy based on their own local state, resulting in load imbalance or increased rotation, increasing the risk of cargo falling. Traditional solutions mostly rely on remote operators for manual intervention, and the delayed response time makes it difficult to meet the timeliness requirements of the high-dynamic environment in the air. In addition, control island effects are more likely to occur in scenarios where communication is interrupted or multi-machine collaborative links fail, making it difficult to ensure the stability and safety of the lifting task. Therefore, there is an urgent need for a collaborative control method to quickly handle sudden lifting failures. Summary of the invention

[0005] The present application provides a method, device, equipment and medium for collaborative lifting of large-load drones, which can quickly handle sudden failures during the collaborative lifting of multiple drones.

[0006] In a first aspect of the present application, a method for collaborative lifting of a large-load UAV is provided, the method being applied to a virtual pilot UAV in a UAV cluster, the method comprising:

[0007] Obtain the node positions and load distribution data of each edge node, where the virtual pilot UAV is the central node of the UAV cluster, and the UAVs in the UAV cluster except the virtual pilot UAV are edge nodes;

[0008] Calculate the total load data of the loads carried by multiple edge nodes based on the load distribution data;

[0009] Based on the total load data and the node positions, determine the overall movement trajectory of the UAV cluster, and send the overall movement trajectory to the edge nodes, so that the edge nodes can find the required reference pulling force value and node movement trajectory based on the ant colony algorithm;

[0010] During flight, if it is determined that the sling of the first UAV among multiple UAVs breaks, according to a predetermined load redistribution scheme, synchronously distribute the pulling force to each second UAV, where the second UAV is the UAV in the UAV cluster except the first UAV;

[0011] Set the first UAV as the proxy central node, and receive the updated movement trajectory of the UAV cluster from the proxy central node.

[0012] On the basis of the above technical solutions, preferably, the calculating the total load data of the loads carried by multiple edge nodes based on the load distribution data specifically includes:

[0013] Obtain the load distribution data calculated by each edge node, where the edge node obtains the sling pulling force, its own vertical acceleration and pitch and roll angles in real time through the tension sensor, accelerometer and attitude sensor carried on it, calculates the vertical component force borne by itself, and then corrects the inertial change of the suspension point based on the inertial acceleration measured by itself, calculates the effective gravitational acceleration correction factor under unit pulling force, and obtains the corrected load distribution data;

[0014] Using all load distribution data as weights, calculate the cargo center of gravity position according to the multiple node positions;

[0015] Determine the pulling force action vector of the edge node according to the direction vector formed between the node position and the cargo center of gravity position;

[0016] Project the pulling force action vector in the direction of gravity to obtain the action component of each load distribution data in the vertical direction;

[0017] According to the acceleration data of the UAV cluster, perform inertial correction on the action component to obtain the corrected action component;

[0018] Weightedly fuse the correction action components according to the relative distance between the node position and the center-of-gravity position of the goods to obtain the total load data.

[0019] Based on the above technical solutions, preferably, determining the overall motion trajectory of the UAV cluster based on the total load data and the node positions specifically includes:

[0020] Construct a suspension point spatial distribution model of the UAV cluster by combining the total load data and the node positions;

[0021] Based on the suspension point spatial distribution model and in combination with the change rate and instantaneous acceleration data of the center-of-gravity position of the goods, establish a center-of-gravity state vector of the goods, and then predict the motion trend of the center-of-gravity of the goods;

[0022] On the basis of the motion trend of the center-of-gravity of the goods, introduce mass distribution constraints, tension balance constraints, and attitude consistency constraints to construct a load-node structure coupling model, thereby establishing a dynamic mapping relationship between the motion trend of the center-of-gravity of the goods and the flight path of the UAV cluster;

[0023] Introduce the multi-body system dynamics modeling method, regard the goods as a concentrated mass body, and connect each of the edge nodes through flexible cables to form a spatial tension network structure, and establish the dynamic control equation of the whole UAV cluster relying on the tension balance constraint and the angular momentum conservation constraint;

[0024] According to the mission track input and the motion trend of the center-of-gravity of the goods, use the trajectory optimal control algorithm to solve the dynamic control equation to form the desired three-dimensional path of the center-of-gravity of the goods;

[0025] Map the desired three-dimensional path to the flight path of the UAV cluster based on the load-node structure coupling model to obtain the overall motion trajectory.

[0026] Based on the above technical solutions, preferably, sending the overall motion trajectory to the edge nodes so that the edge nodes can find the required reference tension value and the node motion trajectory based on the ant colony algorithm specifically includes:

[0027] Obtain the reference tension value calculated by the edge node based on the overall motion trajectory and the node motion trajectory. Among them, the edge node constructs a local path search graph, which consists of the current position of the node, the adjacent path state, the dynamic attitude limit, and the cable geometry boundary, forming a feasible solution graph of the path space. Based on the path search graph, a heuristic search model is constructed. Using the current expected path point in the overall motion trajectory as the target guidance, the transfer probability is controlled by the pheromone intensity and the heuristic function; in each iteration, obtain the satisfaction degree of the path tension combination solution generated by the edge node evaluation ant colony for the tension balance degree and the attitude consistency index of the system, and define the objective function; after multiple rounds of iteration, the edge node selects the node motion trajectory and the reference tension value corresponding to the optimal ant colony path as the execution command for the current control period, and sends the node motion trajectory and the reference tension value to the central node.

[0028] On the basis of the above technical solutions, preferably, during the flight, if it is determined that the first unmanned aerial vehicle (UAV) among multiple UAVs has a suspension rope break, before synchronously distributing the tension to each second UAV according to a predetermined load redistribution plan, the method further includes:

[0029] Construct a suspension point space distribution model of the UAV cluster by combining the total load data and the node positions;

[0030] For each of the edge nodes, construct a virtual failure model for the case of suspension rope break. Without changing the total mass of the goods, respectively remove the load distribution data corresponding to each failed node among multiple edge nodes, and solve a new set of tension distributions for the remaining nodes, so as to minimize the perturbation cost function compared with the suspension point space distribution model while satisfying the conservation of the resultant tension and the angular momentum constraint;

[0031] Use a quadratic programming solver to perform offline enumeration calculation on the perturbation cost function, and generate an optimal redistribution solution corresponding to the suspension rope break for each failed node. Each optimal redistribution solution corresponds to a redistribution configuration for a failure case;

[0032] Obtain the load redistribution plan by aggregating the optimal redistribution solutions corresponding to all the failed nodes.

[0033] On the basis of the above technical solutions, preferably, during the flight, if it is determined that the first UAV among multiple UAVs has a suspension rope break, and synchronously distribute the tension to each second UAV according to a predetermined load redistribution plan, specifically includes:

[0034] Determine the failed node corresponding to the first UAV;

[0035] Invoke the optimal reallocation solution corresponding to the failed node corresponding to the first drone;

[0036] Determine the remaining nodes corresponding to each of the second drones in the optimal reallocation solution, so as to determine the updated tension values and cable direction unit vectors that each of the second drones should bear;

[0037] Broadcast the updated tension values and cable direction unit vectors to each of the second drones respectively.

[0038] Based on the above technical solution, preferably, before obtaining the node positions and load distribution data of each edge node, the method further includes:

[0039] Obtain the real-time state parameters broadcast by each drone in the drone cluster during the takeoff preparation stage, where the real-time state parameters include position information, remaining computing resources, communication signal quality, historical trajectory stability index, and node reliability level;

[0040] Calculate the centrality score of the virtual leading drone according to the real-time state parameters, and broadcast the centrality score of the virtual leading drone;

[0041] After completing the omnidirectional score synchronization and each drone independently compares the magnitudes of the centrality scores of multiple drones, if it is determined that the centrality score of the virtual leading drone is the highest, then determine the virtual leading drone as the central node.

[0042] In the second aspect of the present application, there is provided a large-load drone collaborative lifting device, which is used to execute a large-load drone collaborative lifting method as described in any one of the above, and the device is arranged on the virtual leading drone in the drone cluster. The device includes an acquisition module, a processing module, and an output module, where:

[0043] The acquisition module is used to acquire the node positions and load distribution data of each edge node, where the virtual leading drone is the central node of the drone cluster, and the drones in the drone cluster except the virtual leading drone are edge nodes;

[0044] The processing module is used to calculate the total load data of the loads lifted by multiple edge nodes based on the load distribution data;

[0045] The processing module is used to determine the overall movement trajectory of the drone cluster based on the total load data and the node positions, and send the overall movement trajectory to the edge nodes, so that the edge nodes can find the required reference tension values and node movement trajectories based on the ant colony algorithm;

[0046] The output module is configured to, during flight, if it is determined that the suspension rope of the first unmanned aerial vehicle (UAV) among multiple UAVs breaks, synchronously distribute the pulling force to each second UAV according to a predetermined load redistribution scheme, where the second UAVs are the UAVs in the UAV cluster other than the first UAV;

[0047] The output module is configured to set the first UAV as a proxy central node and receive an updated movement trajectory of the UAV cluster from the proxy central node.

[0048] In a third aspect of the present application, an electronic device is provided, including a processor, a memory, a user interface, and a network interface. The memory is used to store instructions. Both the user interface and the network interface are used to communicate with other devices. The processor is used to execute the instructions stored in the memory so that the electronic device executes the method described in any one of the above.

[0049] In a fourth aspect of the present application, a computer-readable storage medium is provided. The computer-readable storage medium stores instructions that, when executed, execute the method described in any one of the above.

[0050] In summary, one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages:

[0051] 1. By introducing a virtual leading UAV as the central node in the UAV cluster, combining the real-time load data and spatial positions of the edge nodes to construct a global force model, and adopting a pre-calculated load redistribution scheme and a local path optimization mechanism based on the ant colony algorithm, the present application realizes the efficient control of cluster collaborative hoisting. During flight, once it is detected that the suspension rope of a certain UAV breaks, the corresponding optimal reallocation solution can be immediately called, and the updated tension value and direction information are synchronized to all remaining edge nodes through the central node, enabling them to quickly complete tension adjustment and attitude correction. At the same time, the failed node is converted into a proxy central node, enabling it to undertake trajectory update and control tasks without participating in physical hoisting, forming a fault-tolerant mechanism of logical control redundancy and physical load decoupling, thereby achieving a millisecond-level response to sudden failures and an immediate recovery of full stability.

[0052] 2. By constructing a multi-node load perception model based on the fusion calculation of a tension sensor, an accelerometer, and an attitude sensor, and combining the relationship between the spatial position and the center-of-gravity offset, this application comprehensively introduces the projection of the pulling force direction and the dynamic inertia correction mechanism, realizing a high-precision real-time estimation of the total load during the collaborative lifting of a UAV swarm. It effectively overcomes the measurement errors caused by attitude disturbances, non-vertical pulling forces, or dynamic accelerations in traditional methods, improving the robustness and reliability of the total load estimation under asymmetric forces and dynamic flight changes, and providing an accurate mechanical basis and dynamic reference for subsequent path planning, tension coordination, and fault adaptive control.

[0053] 3. By fusing the total load data and the node spatial distribution, this application constructs an accurate suspension point spatial distribution model, and combines the predicted movement trend of the cargo centroid in real time, introducing multiple physical constraints such as mass distribution, tension balance, and attitude consistency, to establish a dynamic mapping relationship between the load and the nodes. Through the multi-body modeling method and the trajectory optimal control algorithm, it realizes the dynamic coordination and optimal matching between the desired path of the cargo and the flight path of the UAV swarm, thereby generating an overall motion trajectory that conforms to the dynamic constraints and has high coordination. It improves the path accuracy, attitude consistency, and tension coordination of the swarm during the collaborative lifting process, and enhances the dynamic stability and task execution ability under complex spatial paths. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 is a schematic flowchart of a method for collaborative lifting of large-load UAVs disclosed in an embodiment of this application;

[0055] Figure 2 is a schematic block diagram of a device for collaborative lifting of large-load UAVs disclosed in an embodiment of this application;

[0056] Figure 3 is a schematic structural diagram of an electronic device disclosed in an embodiment of this application.

[0057] Description of the reference numerals: 201, acquisition module; 202, processing module; 203, output module; 301, processor; 302, communication bus; 303, user interface; 304, network interface; 305, memory. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0058] In order to enable those skilled in the art to better understand the technical solutions in this specification, the following will clearly and completely describe the technical solutions in the embodiments of this specification in conjunction with the accompanying drawings in the embodiments of this specification. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments.

[0059] In the description of the embodiments of the present application, words such as "for example" or "for illustration" are used to give examples, illustrations or explanations. Any embodiment or design solution described as "for example" or "for illustration" in the embodiments of the present application should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Rather, the use of words such as "for example" or "for illustration" is intended to present relevant concepts in a specific manner.

[0060] In the description of the embodiments of the present application, the term "plural" means two or more. For example, plural systems refer to two or more systems, and plural screen terminals refer to two or more screen terminals. In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. The terms "include", "comprise", "have" and their variants all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0061] For large unmanned aerial vehicles (UAVs) with a take-off mass greater than 150 kg, that is, large payload UAVs, the cooperative lifting technology of large payload UAVs realizes the air transportation of large-size or heavy-load goods through cluster control and load distribution, and relies on high-precision perception, multi-source fusion and distributed control to complete stable flight and load balance. Although progress has been made in path planning and disturbance rejection control, in the face of non-structural sudden failures such as the breakage of the lifting rope, existing systems generally lack real-time perception and adaptive response mechanisms, and it is difficult to reconstruct the tension distribution strategy in a timely manner, which easily leads to load imbalance and control failure. There is an urgent need for an intelligent cooperative control method with fast fault tolerance ability to improve system stability and mission safety.

[0062] This embodiment discloses a cooperative lifting method for large payload UAVs, referring to Figure 1 , including the following steps S110 - S150:

[0063] S110, obtain the node positions and load distribution data of each edge node.

[0064] A cooperative lifting method for large payload UAVs disclosed in the embodiments of the present application is applied to a virtual leader UAV in a UAV cluster. The virtual leader UAV serves as the central node of the cluster in the multi-UAV cooperative lifting system, is responsible for fusing the position information and load distribution data uploaded by each edge node, uniformly planning the overall motion trajectory and generating target control instructions, and at the same time scheduling the load distribution strategy and quickly reconstructing the control architecture in the event of sudden failures such as the breakage of the lifting rope, ensuring that the system maintains flight stability and mission continuity in a high-dynamic environment, and is the core execution unit for cluster cooperative control and global mission coordination.

[0065] In a possible implementation manner, before obtaining the node positions of each edge node and the load distribution data measured by the sensors in real time, the method further includes: obtaining the real-time state parameters broadcast by each unmanned aerial vehicle (UAV) in the UAV cluster during the takeoff preparation phase, where the real-time state parameters include position information, remaining computing resources, communication signal quality, historical trajectory stability index, and node reliability level; calculating the centrality score of the virtual leading UAV according to the real-time state parameters, and broadcasting the centrality score of the virtual leading UAV; after completing the omnidirectional score synchronization and each UAV independently compares the magnitudes of the centrality scores of multiple UAVs, if it is determined that the centrality score of the virtual leading UAV is the highest, then determine the virtual leading UAV as the central node.

[0066] When implementing this technical solution, first, during the takeoff preparation phase, each UAV in the UAV cluster actively broadcasts its current real-time state parameters, which include its position information in the three-dimensional space, the current remaining computing resource ratio, such as CPU or GPU utilization rate, the communication signal quality with other nodes, such as RSSI value or packet loss rate, the historical trajectory stability index, obtained by calculating the position fluctuation amplitude and speed mutation frequency within a short period of time, and the node reliability level statistically obtained based on past task records, such as failure rate or response success rate. These parameters are periodically broadcast by each node and synchronized to all members in the cluster through a low-latency self-organizing network.

[0067] Subsequently, each UAV independently calculates the centrality score of the virtual leading UAV candidate locally using all the obtained node state parameters based on a preset weighted model. The scoring function combines multiple dimensional indicators to reflect the resource scheduling ability, communication hub potential, and operation stability of the node in the cluster. The specific form is to normalize each parameter and then accumulate it according to the weighted coefficient to ensure comparability between different dimensions and adjust the weight distribution according to task requirements. For example, when the weights of communication quality and computing resources are relatively high, a node with fast response and strong connection ability is preferentially selected as the control core. After each node completes the score calculation, it broadcasts its centrality score for the virtual leading candidate node.

[0068] After all the scoring data is synchronized in the cluster, each drone node independently compares the centrality scores of all candidate nodes based on the broadcast scoring results and selects the one with the highest score using a deterministic selection mechanism. To avoid conflicts caused by tied scores, a priority adjudication rule can be further set using the node ID or timestamp when the scoring results are the same, ensuring that there is always a unique and consistent control center in the cluster at any given time. When the score of the finally determined virtual lead drone is confirmed to be the highest globally and meets centrality constraints such as being centered in position or having the strongest communication accessibility, it is automatically set as the central node for this round of tasks, given the dominant right of path planning, task scheduling, and control broadcasting. The remaining nodes act as edge nodes and accept the instructions of this central node to perform collaborative operations, forming a hierarchical multi-drone control architecture to ensure the stable and efficient operation of subsequent collaborative control.

[0069] In this collaborative lifting method, the virtual lead drone, as the central node of the drone cluster, is responsible for obtaining its position information and the load distribution data measured by the tension sensor, accelerometer, and attitude module from all edge nodes in real time. Specifically, each edge node determines its relative pose with respect to the center of gravity of the cargo based on its three-dimensional spatial position, measures the tension of its suspension rope and its own attitude angle, extracts the vertical component of the tension after local preprocessing to form an estimated data packet of its own load share, and simultaneously sends its current precise position to the central node through low-latency wireless communication. After receiving the data from all edge nodes, the central node completes the reconstruction of the cluster's spatial distribution structure and the overall perception of the tension field, providing an accurate data basis for subsequent motion trajectory generation, tension redistribution, and emergency fault response.

[0070] S120, calculate the total load data of the loads lifted by multiple edge nodes based on the load distribution data.

[0071] In a possible implementation, the total load data of the loads lifted by multiple edge nodes is calculated based on the load distribution data, which specifically includes: obtaining the load distribution data calculated by each edge node. Among them, the edge node obtains the hoisting rope tension, its own vertical acceleration, and pitch and roll angles in real time through the mounted tension sensor, accelerometer, and attitude sensor, calculates the vertical component of the force it bears, then corrects the inertial change of the hanging point based on the inertial acceleration measured by itself, calculates the effective gravitational acceleration correction factor under unit tension, and obtains the corrected load distribution data; using all the load distribution data as weights, calculates the center of gravity position of the goods according to the positions of multiple nodes; determines the tension action vector of the edge node according to the direction vector formed between the node position and the center of gravity position of the goods; projects the tension action vector in the direction of gravity to obtain the action component of each load distribution data in the vertical direction; performs inertial correction on the action component according to the acceleration data of the UAV cluster to obtain the corrected action component; performs weighted fusion on the corrected action component according to the relative distance between the node position and the center of gravity position of the goods to obtain the total load data.

[0072] Specifically, first, each edge node measures the hoisting rope tension data in real time based on the mounted tension sensor, and at the same time uses the accelerometer to obtain its own vertical acceleration information along the direction of gravity, and combines the pitch angle and roll angle provided by the attitude sensor to calculate the vertical force component of the hoisting rope tension borne by the edge node through the attitude compensation algorithm. For each i-th edge node, its tension measurement value is T i , the pitch angle is θ i , the roll angle is φ i , and the vertical component of its tension is expressed as:

[0073] F zi = T i ·cos(θ i )·cos(φ i )

[0074] Subsequently, the edge node further uses the inertial acceleration data obtained by its inertial measurement unit to correct the inertial change generated by dynamic disturbance at its hanging point, and calculates the corrected gravitational acceleration factor corresponding to unit tension. Let the vertical acceleration measured by the i-th edge node be a zi , then its corrected unit gravitational acceleration factor is:

[0075]

[0076] where g is the standard gravitational acceleration constant.

[0077] To correct the physical meaning of the vertical component of force and finally obtain the payload allocation data of the current edge node after inertial compensation under dynamic conditions, the estimated value of the load sharing of the i-th edge node is:

[0078]

[0079] This payload allocation data is used to participate in the global load estimation. Then, the central node, as the central node of the UAV cluster, receives the payload allocation data after inertial correction uploaded by all edge nodes, and uses this payload allocation data as a weighting factor. The weighting factor is combined with the actual coordinate positions of each edge node in three-dimensional space, and the spatial position of the center of gravity of the goods at the current moment is calculated based on the load-weighted geometric centroid model. Let the spatial position vector of the i-th edge node be r i =(x i , y i , z i ), then the center of gravity position vector r c is expressed as:

[0080]

[0081] This spatial position of the center of gravity is regarded as the point of action of the resultant force of the tensions of each edge node in space. Then, the central node constructs a direction vector from its spatial position to the spatial position of the center of gravity of the goods for each edge node. This direction vector characterizes the tension direction of this edge node, that is, the directional feature of its actual force application path in space. Let the direction vector d i point from the i-th edge node to the center of gravity r c , then there is:

[0082]

[0083] Subsequently, the central node performs a vector projection operation on the tension direction vector of each of the above edge nodes, projects this tension direction vector onto the unit vector defined by the gravity direction, and obtains the projection component of the tension vector of this edge node in the gravity direction, that is, the actual support force in the vertical direction. Let the unit vector of the gravity direction be e z =(0, 0, -1), then the projection component of the tension direction of the i-th edge node is:

[0084]

[0085] Since the overall motion state of the entire UAV cluster and the goods being lifted may change during flight, the central node further obtains the reference acceleration data of the overall cluster, performs inertial correction on the above projection components, and eliminates the inertial interference caused by the accelerated motion of the system, so as to obtain the inertial correction acting component of each edge node. This inertial correction acting component reflects its actual mechanical contribution to counteracting gravity. Introduce the measured value a of the overall centroid acceleration of the goods in the vertical direction c , then the corrected effective acting component is:[[]]

[0086]

[0087] Finally, the central node calculates the spatial penalty weight factor according to the relative Euclidean distance between each edge node and the spatial position of the goods center of gravity. This spatial penalty weight factor is used to adjust the acting credibility of the edge nodes. Let R i = ||r i - r c ||, the maximum relative distance is R max , then the distance weight of the i-th edge node is:[[]]

[0088]

[0089] For edge nodes that are far from the center of gravity position, their inertial correction acting components will be given a lower fusion weight to avoid the influence of the asymmetric moment interference caused by too long a tension arm on the total load estimation. After the inertial correction acting components of all edge nodes are weighted and fused, the total load data W of the current UAV cluster for the lifted target object is obtained total :[[]]

[0090]

[0091] This total load data is used as an input variable in the global system control model to support the implementation of subsequent control tasks such as cooperative trajectory generation, reference tension distribution, and adaptive reconfiguration for sudden failures.[[]]

[0092] S130. Based on the total load data and the node positions, determine the overall motion trajectory of the UAV cluster, and send the overall motion trajectory to the edge nodes.[[]]

[0093] In a possible implementation, based on the total load data and the node positions, the overall motion trajectory of the UAV cluster is determined, which specifically includes: constructing a suspension point spatial distribution model of the UAV cluster by combining the total load data and the node positions; based on the suspension point spatial distribution model, and combining the change rate and instantaneous acceleration data of the cargo center of gravity position, establishing a cargo centroid state vector, and then predicting the motion trend of the cargo centroid; on the basis of the cargo centroid motion trend, introducing mass distribution constraints, tension balance constraints and attitude consistency constraints, constructing a load-node structure coupling model, so as to establish a dynamic mapping relationship between the cargo centroid motion trend and the flight path of the UAV cluster; introducing the multi-body system dynamics modeling method, regarding the cargo as a concentrated mass body, connecting each edge node through flexible cables to form a spatial tension network structure, and relying on the tension balance constraint and the angular momentum conservation constraint to establish the overall dynamic control equation of the UAV cluster; according to the task track input and the cargo centroid motion trend, using the trajectory optimal control algorithm to solve the dynamic control equation to form the expected three-dimensional path of the cargo centroid; based on the load-node structure coupling model, mapping the expected three-dimensional path to the flight path of the UAV cluster to obtain the overall motion trajectory.

[0094] Specifically, first, the virtual leader UAV constructs a suspension point spatial distribution model according to the calculated total load data and the three-dimensional spatial positions of each edge node. Let the three-dimensional position vector of the i-th edge node be

[0095]

[0096] This suspension point spatial distribution model uses the position vector of the edge node as the tension application point, and combines the spatial vector difference between each edge node and the cargo center of gravity to define the cable connection configuration. The total load data is W total , and the tension distribution data is T i . According to the tension weighted calculation of the cargo center of gravity spatial position:

[0097]

[0098] Define the cable vector as:

[0099] l i =r c -r i

[0100] The set {l i} constitutes the suspension point spatial distribution model.

[0101] Thus, a spatial mechanical connection topology structure of the UAV cluster to the cargo is established, providing a geometric basis for subsequent coupling modeling.

[0102] Next, a velocity vector is formed based on the change rate of the cargo centroid position over time, and combined with the acceleration vector provided by the inertial measurement unit, a centroid state vector including position, velocity, and acceleration is constructed. Let the position of the cargo centroid be r c (t), the velocity be v c (t), the acceleration be a c (t), and the state vector be:

[0103]

[0104] By continuously sampling and estimating the centroid state vector, the state prediction algorithm is used to predict the centroid motion trend, which characterizes the future displacement trajectory and dynamic response trend of the cargo. The centroid motion trend is represented by the following state update formula:

[0105] x c (t + Δt) = A · x c (t) + B · u(t)

[0106] where A is the state transition matrix, B is the control input matrix, u(t) is the input disturbance vector, and the state vector can be predicted and estimated by the extended Kalman filter.

[0107] Based on this centroid motion trend, mass distribution constraints, tension balance constraints, and attitude consistency constraints are introduced to limit the upper limit of the thrust output of the edge nodes, the balance and coordination relationship of the tensions of the cables inside, and the synchronization of the attitudes of all edge nodes, respectively. Among them, the mass distribution constraint is expressed as follows:

[0108]

[0109] The tension balance constraint is expressed as follows:

[0110]

[0111] The attitude consistency constraint is expressed as follows:

[0112]

[0113] where, R i is the attitude matrix of the i-th edge node, and ∈ sync is the consistency threshold.

[0114] Under these triple constraint conditions, a load-node structure coupling model is established, which describes the spatial mechanical coupling relationship between the change of the cargo centroid state and the flight state of the UAV, and establishes the mapping logic between the cluster power output and the cargo response.

[0115] Subsequently, a multi-body dynamics modeling method is introduced. The cargo is modeled as a concentrated mass body, all cables are modeled as flexible tensile constraints, and all edge nodes are modeled as spatial moving rigid bodies. In this multi-body dynamics model, a tensile force balance constraint is established to ensure that the resultant force of all cable tension vectors cancels out the gravity and inertial load. At the same time, an angular momentum conservation constraint is introduced to maintain rotational stability in three-dimensional space, forming a cluster dynamics control equation. With the mass of the cargo being m c , a force balance equation is constructed by introducing the tension effect:

[0116]

[0117] The angular momentum conservation equation is expressed as:

[0118]

[0119] where I c is the inertia tensor of the cargo's rotation, and ω c is the angular velocity vector. This dynamics control equation couples the node tension, pose, spatial position, and the response of the cargo's centroid.

[0120] According to the input information of the mission trajectory and the predicted centroid motion trend, a trajectory optimal control algorithm is used to solve the above dynamics control equation. The desired path is defined as The objective is to minimize the following performance index function:

[0121]

[0122] where ρ i is the cost coefficient of tension change, is the desired tension curve. The tension trajectory and the centroid path are obtained by solving through optimal control algorithms such as LQR, NMPC, or the Pontryagin minimum principle. This trajectory optimal control algorithm is based on a constraint optimization and nonlinear trajectory iterative correction mechanism. Under the premise of considering tension constraints, dynamic limits, and response time, it outputs the desired centroid path, which is the optimal three-dimensional trajectory prediction of the cargo under the current dynamic model.

[0123] Finally, based on the load-node structure coupling model, this desired centroid path is converted into the overall motion trajectory of the UAV cluster through spatial mechanics mapping. The centroid path r c (t) is mapped to the overall flight path of the UAV cluster through the mechanics mapping relationship:

[0124] r i (t) = r c (t) - l i ·u i (t)

[0125] where, is the unit vector of the cable direction, l i is the cable length constant. This overall motion trajectory ensures that the overall behavior of the cluster is highly consistent with the centroid path of the cargo, thereby achieving stable cooperative control of the lifting operation in a complex dynamic environment.

[0126] Furthermore, after generating the overall motion trajectory, the virtual leader UAV broadcasts it as a three-dimensional desired path vector set to all edge nodes through the intra-cluster communication link. This overall motion trajectory consists of a sequence of desired centroid spatial positions at consecutive time steps, the dynamic state vector of the cargo, and attitude evolution information, serving as the path reference framework for the global cooperative flight of the cluster. After receiving this overall motion trajectory, each edge node does not directly execute it but instead uses it as the target input for path search constraints and tension control optimization, initiating a trajectory-tension collaborative solution process based on the ant colony algorithm.

[0127] Specifically, each edge node constructs a local path search graph, which is composed of the current position of the node, adjacent path states, dynamic attitude constraints, and cable geometric boundaries, forming a feasible solution graph of the path space. The ant colony algorithm constructs a heuristic search model based on the path search graph, where each "ant" represents a feasible combination solution of the node motion trajectory and tension output. The path search process is guided by the current desired path point in the overall motion trajectory, and the transition probability is controlled by the pheromone intensity τ and the heuristic function η:

[0128]

[0129] where is the set of actionable actions in the current state, and η ij (t) is composed of the distance between the current position and the overall desired path, the attitude change rate, and the tension balance degree, reflecting the degree of fit of this path to the global goal.

[0130] In each iteration, the edge node evaluates the satisfaction degree of the path-tension combination solution generated by the ant colony for the system's tension balance degree and attitude consistency index, and defines the objective function:

[0131]

[0132] where is derived from the mapping of the overall motion trajectory, is the local tension target, and ΔR i (t) is the attitude change amount. The higher the quality of the path-tension pair, the higher the intensity it obtains in the ant colony pheromone update, enhancing the probability advantage of this path in subsequent searches.

[0133] After multiple rounds of iteration, the edge nodes select the flight trajectory corresponding to the optimal ant colony path and the tension output as the execution command for the current control cycle, and share the results with adjacent edge nodes for synchronization to ensure the overall coordination of the tension distribution within the cluster. This distributed path optimization mechanism utilizes the global search ability and local feedback structure of the ant colony algorithm, enabling each edge node to autonomously find the optimal reference pulling force value and the adapted node motion trajectory while following the overall motion trajectory, thereby achieving the adaptive tension control and structural stability maintenance of the cooperative lifting system.

[0134] S140. During flight, if it is determined that the suspension rope of the first unmanned aerial vehicle (UAV) among multiple UAVs breaks, according to a predetermined load reallocation scheme, synchronously distribute the pulling force to each second UAV.

[0135] In a possible implementation manner, before synchronously distributing the pulling force to each second UAV according to a predetermined load reallocation scheme when it is determined that the suspension rope of the first UAV among multiple UAVs breaks during flight, the method further includes: constructing a suspension point spatial distribution model of the UAV cluster by combining the total load data and the node positions; constructing a virtual failure model for the case of suspension rope breakage for each edge node, and without changing the total mass of the goods, respectively removing the load distribution data corresponding to each failed node among multiple edge nodes, and solving a new set of tension distributions for the remaining nodes, so as to minimize the perturbation cost function compared with the suspension point spatial distribution model while satisfying the conservation of the resultant tension and the angular momentum constraint; using a quadratic programming solver to perform offline enumeration calculations on the perturbation cost function, generating an optimal reallocation solution corresponding to the suspension rope breakage for each failed node, and each optimal reallocation solution corresponding to a reallocation configuration for a failure case; and obtaining the load reallocation scheme by aggregating the optimal reallocation solutions corresponding to all failed nodes.

[0136] Specifically, first, the central node constructs a suspension point spatial distribution model of the UAV cluster based on the currently obtained total load data and the spatial position vectors of each edge node. This suspension point spatial distribution model uses each edge node as the pulling force application point, takes the center of gravity of the goods as the center point, defines each cable connection vector, forms a set of tension action directions, and constructs a cable mechanical topology structure on this basis to represent the pulling force transmission path, direction distribution, and node configuration in the current situation, serving as a steady-state force-bearing reference.

[0137] Subsequently, for each edge node, a corresponding virtual failure model of sling breakage is constructed. Based on the "single-node breakage hypothesis", this virtual failure model removes the tension data and connection vectors of this node from the original load distribution model while keeping the total mass of the goods and the external disturbing force unchanged, retains the effective spatial configuration and historical tension information of the remaining nodes, and re-establishes the force equations of the remaining clusters according to the new node configuration, including the tension resultant force balance equation and the angular momentum conservation equation, as the boundary constraints for solving the reallocation strategy.

[0138] Taking the original sling point spatial distribution model as a reference, the deviation degree of the post-failure state from the original state is measured by defining a perturbation cost function. This perturbation cost function includes a node position perturbation term, a tension reconstruction offset term, and a resultant force direction offset term, and its mathematical expression is:

[0139]

[0140] where J is the perturbation cost function, which is the objective function of this optimization, representing the overall perturbation degree introduced by the tension reallocation state after the sling breakage compared to the original state. The goal of the optimization is to minimize the value of this function while satisfying the mechanical constraints. i ∈ S ′ is the set of edge node indices, representing the numbers of all edge nodes except the failed node in the current assumed breakage scenario. is the tension allocation value recalculated for edge node i in the case of node failure, that is, the updated tensile force that this node needs to bear. is the original tension allocation value before node failure, representing the historical tensile force task of this node under steady-state working conditions. This term represents the change amplitude of the node tension value, measuring the degree to which each edge node deviates from its original tension after the tension is reallocated, and reflecting the perturbation degree to the original mechanical state. u i is the unit vector of the tensile force direction of edge node i, representing the direction from this node position to the center of gravity of the goods, and is used to project the tension vector into space for resultant force calculation. is the resultant force vector of all remaining node tension vectors in three-dimensional space after the current tension reallocation. W total is the estimated total weight vector of the goods, that is, the resultant force direction and magnitude that should be generated during hoisting under ideal conditions, and is usually set as a constant vector with the gravity direction downward. γ is the weight coefficient of the resultant force constraint term, used to adjust the attention degree to the overall force balance during tension reconstruction. The larger the value, the more the optimization result tends to prioritize satisfying the resultant force conservation condition. represents the deviation between the current reallocated tension resultant force and the target total load, constraining the synthetic tension provided by the UAV cluster to be consistent with the actual force on the goods, and ensuring that no overall imbalance will occur. τ newThe angular momentum vector after re - distributing the tension, defined as the sum of the torques of each remaining node with respect to the center of gravity of the cargo. τ orig The angular momentum vector in the original stable state, which should ideally be close to zero, indicating no spin and no torsion. ||τ new -τ orig || 2 It represents the degree of perturbation in the rotational equilibrium after tension reconstruction, reflecting whether it will cause the cargo to rotate or the attitude to become unstable. δ is the weight coefficient of the angular momentum constraint term, used to control the degree of emphasis of the optimization algorithm on attitude stability. The larger the weight, the more inclined to keep the original angular momentum unchanged, ensuring the minimum rotational perturbation.

[0141] For the above - mentioned perturbation cost function, a quadratic programming solver is used for constrained optimization calculation, setting non - negative tension and maximum thrust boundary conditions to ensure the feasibility and physical validity of the solution. To improve the timeliness of fault response, this solution process is enumerated and pre - processed as an offline calculation task before flight. For each edge node, a failure scenario is independently constructed, and the corresponding optimal tension re - distribution solution is solved in turn to form a set of re - distribution configuration sets covering all edge nodes.

[0142] Finally, the optimal tension re - distribution solutions corresponding to the failure cases of all edge nodes are unified and integrated to form a load re - distribution plan. This load re - distribution plan is stored as a mapping table in the control module of the central node. Once a suspension rope of an edge node breaks during the actual flight, the pre - calculated re - distribution solution can be immediately called, skipping the real - time solution process, to achieve highly responsive and highly robust mechanical adaptive control.

[0143] In a possible implementation, during flight, if it is determined that the suspension rope of the first unmanned aerial vehicle (UAV) among multiple UAVs breaks, according to the predetermined load re - distribution plan, the pulling force is synchronously distributed to each second UAV, specifically including: determining the failure node corresponding to the first UAV; calling the optimal re - distribution solution corresponding to the failure node of the first UAV; determining the remaining nodes corresponding to each second UAV in the optimal re - distribution solution, so as to determine the updated tension values and the unit vectors of the cable directions that each second UAV should bear; broadcasting the updated tension values and the unit vectors of the cable directions to each second UAV respectively.

[0144] Specifically, first, the central node or each edge node confirms that the suspension rope of the first UAV breaks by real - time monitoring the tension sensor data, vertical acceleration, and flight attitude information of each UAV in the UAV cluster, combined with indicators such as the tension value suddenly dropping to near zero, the sudden change in the acceleration direction, and the attitude instability. Then, the first UAV is marked as a failure node in the cluster control architecture, and its breaking time and the last stable state parameters are recorded as the event source for triggering the load re - distribution mechanism.

[0145] After determining the failed node, the central node immediately calls the optimal reallocation solution corresponding to the failed node from the offline reallocation solution library stored locally. This optimal reallocation solution is the minimum perturbation configuration generated by enumeration through a quadratic programming solver based on the suspension point space distribution model, the tension coupling structure, and the perturbation cost function during the initialization phase of the flight mission. It includes the new tension target values that each second unmanned aerial vehicle (UAV) should bear in the current fracture situation and the updated cable direction unit vectors.

[0146] According to this optimal reallocation solution, the system identifies all the remaining edge nodes as the current set of second UAVs and extracts the tension instruction parameters that each second UAV needs to update and execute, specifically including the target tension value and the cable direction unit vector. This unit vector is calculated based on the newly constructed mechanical balance point after the rope break and the spatial positions of the remaining nodes, and is used to guide the attitude adjustment and direction control of each edge node, so that the acting direction of the pulling force is consistent with the new center of force.

[0147] Subsequently, the central node synchronously sends the sub-node data packets containing the tension and direction instructions to each second UAV in a broadcast manner. After receiving them, each edge node starts the local controller. The local controller adjusts the propeller speed through a feedback gain adjustment control algorithm based on the error between the current tension measurement value and the target tension value, and then realizes the dynamic adjustment of the output vertical force vector, so that the actual tension gradually approaches the set value.

[0148] At the same time, each second UAV uses the attitude control module to finely adjust the flight attitude, and real-time corrects its own spatial position and orientation according to the received cable direction unit vector, so that the suspension rope direction vector is consistent with the newly calculated expected direction in the three-dimensional space, thereby restoring the tension balance in the physical space and suppressing the system rotation trend, and finally realizing the rapid recovery of the system stability and the seamless continuation of the lifting task. The whole process is characterized by the combination of a distributed control architecture and central guidance, ensuring the unity of response timeliness and control coordination.

[0149] S150, set the first UAV as the proxy central node and receive the updated motion trajectory of the UAV cluster for the proxy central node.

[0150] After confirming that the first UAV has a broken suspension rope and completing the failure identification, switch this first UAV to the proxy central node, and its role changes from a physical load participant to a logical control coordinator. In the specific implementation process, the virtual leading UAV judges whether the communication link quality, the remaining computing resources, and the spatial position stability meet the requirements of the central node responsibilities according to the state parameters recorded before the rope break. If they are met, it immediately issues a role switching instruction and broadcasts a status notice that this node has become the proxy central node within the cluster.

[0151] After the first UAV takes over the central node function, based on the resource redundancy that it no longer bears the actual load, it uses its own inertial measurement unit and the tension and trajectory data uploaded by the surrounding edge nodes to reconstruct the current cargo center-of-gravity state and the suspension point spatial structure in real time. Combining the task input and the dynamic mapping model, it regenerates the updated motion trajectory of the UAV cluster. This updated motion trajectory includes the expected path of the cargo centroid, the node attitude constraints, and the tension scheduling reference. After being generated, it is broadcast by the proxy central node to each edge node through a wide-area communication link to ensure the path control continuity and collaborative stability after the structural mutation during the hoisting.

[0152] This embodiment also discloses a large-load UAV collaborative hoisting device. The device is used to execute a large-load UAV collaborative hoisting method as described in any one of the above, and the device is set on the virtual pilot UAV in the UAV cluster. Referring to Figure 2 , the device includes an acquisition module 201, a processing module 202, and an output module 203, where:

[0153] The acquisition module 201 is used to acquire the node positions and load distribution data of each edge node. Among them, the virtual pilot UAV is the central node of the UAV cluster, and the UAVs in the UAV cluster except the virtual pilot UAV are edge nodes.

[0154] The processing module 202 is used to calculate the total load data of the loads hoisted by multiple edge nodes based on the load distribution data.

[0155] The processing module 202 is used to determine the overall motion trajectory of the UAV cluster based on the total load data and the node positions, and send the overall motion trajectory to the edge nodes, so that the edge nodes can find the required reference tension value and the node motion trajectory based on the ant colony algorithm.

[0156] The output module 203 is used to, if it is determined during the flight that the hoisting rope of the first UAV among multiple UAVs breaks, synchronize the tension distribution to each second UAV according to a predetermined load reallocation scheme. The second UAV is the UAV in the UAV cluster except the first UAV.

[0157] The output module 203 is used to set the first UAV as the proxy central node and receive the updated motion trajectory of the UAV cluster from the proxy central node.

[0158] In a possible implementation, an acquisition module 201 is configured to acquire load distribution data calculated by each edge node. The edge node acquires the sling tension, its own vertical acceleration, and pitch and roll angles in real time through the mounted tension sensor, accelerometer, and attitude sensor, calculates the vertical component force borne by itself, then corrects the inertial change of the suspension point based on the inertial acceleration measured by itself, calculates the effective gravitational acceleration correction factor under unit tension, and obtains the corrected load distribution data.

[0159] A processing module 202 is configured to calculate the center of gravity position of the goods based on the positions of multiple nodes with all the load distribution data as weights.

[0160] The processing module 202 is configured to determine the tension action vector of the edge node according to the direction vector formed between the node position and the center of gravity position of the goods.

[0161] The processing module 202 is configured to project the tension action vector in the direction of gravity to obtain the action component of each load distribution data in the vertical direction.

[0162] The processing module 202 is configured to perform inertial correction on the action component according to the acceleration data of the UAV cluster to obtain the corrected action component.

[0163] The processing module 202 is configured to perform weighted fusion on the corrected action component according to the relative distance between the node position and the center of gravity position of the goods to obtain the total load data.

[0164] In a possible implementation, the processing module 202 is configured to construct a suspension point spatial distribution model of the UAV cluster by combining the total load data and the node positions.

[0165] The processing module 202 is configured to establish a center-of-mass state vector of the goods based on the suspension point spatial distribution model and in combination with the change rate and instantaneous acceleration data of the center of gravity position of the goods, and then predict the movement trend of the center of mass of the goods.

[0166] The processing module 202 is configured to introduce mass distribution constraints, tension balance constraints, and attitude consistency constraints based on the movement trend of the center of mass of the goods to construct a load-node structure coupling model, thereby establishing a dynamic mapping relationship between the movement trend of the center of mass of the goods and the flight path of the UAV cluster.

[0167] The processing module 202 is configured to introduce a multi-body system dynamics modeling method, regard the goods as a concentrated mass body, connect each edge node through a flexible cable to form a spatial tension network structure, and establish the overall dynamic control equation of the UAV cluster based on the tension balance constraint and the angular momentum conservation constraint.

[0168] The processing module 202 is configured to solve the dynamic control equation by using a trajectory optimal control algorithm according to the task track input and the cargo centroid motion trend, and form an expected three-dimensional path of the cargo centroid.

[0169] The processing module 202 is configured to map the expected three-dimensional path to the flight path of the UAV cluster based on the load-node structure coupling model to obtain the overall motion trajectory.

[0170] In a possible implementation manner, the acquisition module 201 is configured to acquire a reference tension value and a node motion trajectory calculated by an edge node based on the overall motion trajectory. The edge node constructs a local path search graph, and the path search graph consists of the current position of the node, the adjacent path state, the dynamic attitude limit, and the cable geometric boundary to form a feasible solution graph of the path space. A heuristic search model is constructed based on the path search graph, and the current expected path point in the overall motion trajectory is used as the target guidance to control the transition probability through the pheromone intensity and the heuristic function. In each iteration, the satisfaction degree of the path tension combination solution generated by the edge node evaluation ant colony for the system tension balance degree and the attitude consistency index is obtained, and the objective function is defined. After multiple rounds of iteration, the edge node selects the node motion trajectory and the reference tension value corresponding to the optimal ant colony path as the execution command for the current control period, and sends the node motion trajectory and the reference tension value to the central node.

[0171] In a possible implementation manner, the processing module 202 is configured to construct a suspension point space distribution model of the UAV cluster in combination with the total load data and the node positions.

[0172] The processing module 202 is configured to construct a virtual failure model for the case of sling breakage for each edge node. Without changing the total mass of the cargo, the load distribution data corresponding to each failed node among multiple edge nodes is respectively removed, and a new set of tension distributions is solved for the remaining nodes, so as to minimize the perturbation cost function compared with the suspension point space distribution model while satisfying the conservation of the resultant tension and the angular momentum constraint.

[0173] The processing module 202 is configured to perform offline enumeration calculation on the perturbation cost function by using a quadratic programming solver, and generate an optimal reallocation solution corresponding to the sling breakage for each failed node. Each optimal reallocation solution corresponds to a reallocation configuration for a failure case.

[0174] The processing module 202 is configured to obtain a load reallocation scheme from the set of optimal reallocation solutions corresponding to all failed nodes.

[0175] In a possible implementation manner, the processing module 202 is configured to determine the failed node corresponding to the first UAV.

[0176] The processing module 202 is configured to call the optimal reallocation solution corresponding to the failed node of the first drone.

[0177] The processing module 202 is configured to determine the remaining nodes corresponding to each second drone in the optimal reallocation solution, so as to determine the updated tension value and the cable direction unit vector that each second drone should bear.

[0178] The output module 203 is configured to broadcast the updated tension value and the cable direction unit vector to each second drone respectively.

[0179] In a possible implementation manner, the acquisition module 201 is configured to acquire the real-time state parameters broadcast by each drone in the drone cluster during the takeoff preparation phase. The real-time state parameters include position information, remaining computing resources, communication signal quality, historical trajectory stability index, and node reliability level.

[0180] The output module 203 is configured to calculate the centrality score of the virtual leading drone according to the real-time state parameters and broadcast the centrality score of the virtual leading drone.

[0181] The processing module 202 is configured to compare the centrality scores of multiple drones independently for each drone after the omnidirectional score synchronization is completed. If it is determined that the centrality score of the virtual leading drone is the highest, the virtual leading drone is determined as the central node.

[0182] It should be noted that: when the device provided in the above embodiment realizes its functions, only the above-mentioned division of each functional module is used for illustration. In practical applications, the above functions can be allocated to different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the device and method embodiments provided in the above embodiment belong to the same concept, and the specific implementation process can be seen in the method embodiment, which will not be elaborated here.

[0183] This embodiment also discloses an electronic device. Referring to Figure 3 , the electronic device may include: at least one processor 301, at least one communication bus 302, a user interface 303, a network interface 304, and at least one memory 305.

[0184] Among them, the communication bus 302 is used to realize the connection and communication between these components.

[0185] Among them, the user interface 303 may include a display screen (Display) and a camera (Camera). Optionally, the user interface 303 may further include a standard wired interface and a wireless interface.

[0186] Among them, the network interface 304 may optionally include a standard wired interface, a wireless interface (such as a WI-FI interface).

[0187] Among them, the processor 301 may include one or more processing cores. The processor 301 connects various parts within the entire server through various interfaces and lines, and executes various functions of the server and processes data by running or executing instructions, programs, code sets, or instruction sets stored in the memory 305, and by calling the data stored in the memory 305. Optionally, the processor 301 may be implemented in at least one hardware form of digital signal processing (DSP), field-programmable gate array (FPGA), or programmable logic array (PLA). The processor 301 may integrate one or a combination of several of a central processing unit (CPU), a graphics processing unit (GPU), and a modem, etc. Among them, the CPU mainly processes the operating system, user interface, application programs, etc.; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 301 and may be implemented separately by a single chip.

[0188] Among them, the memory 305 may include random access memory (RAM) and may also include read-only memory. Optionally, the memory includes a non-transitory computer-readable storage medium. The memory 305 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 305 may include a program storage area and a data storage area. Among them, the program storage area may store instructions for implementing the operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned method embodiments, etc.; the data storage area may store the data involved in the above-mentioned method embodiments. The memory 305 may optionally also be at least one storage device located far from the aforementioned processor 301. The memory 305, as a computer storage medium, may include an operating system, a network communication module, a user interface 303 module, and an application program for a method of cooperative hoisting of a large-load unmanned aerial vehicle.

[0189] In Figure 3In the electronic device shown, the user interface 303 is mainly used to provide an interface for the user to input and obtain the data input by the user; and the processor 301 can be used to call the application program stored in the memory 305 for a large-load unmanned aerial vehicle collaborative lifting method. When executed by one or more processors 301, the electronic device is caused to execute the method in one or more of the above embodiments.

[0190] It should be noted that, for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that this application is not limited by the described action sequence, because according to this application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0191] In the above embodiments, the descriptions of the respective embodiments have their own emphases. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0192] In several embodiments provided by this application, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is only a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed mutual coupling, direct coupling, or communication connection can be through some service interfaces. The indirect coupling or communication connection of the device or unit can be in an electrical or other form.

[0193] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place, or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0194] In addition, the functional units in each embodiment of this application can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0195] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a memory 305 and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present application. The aforementioned memory 305 includes various media that can store program codes, such as USB flash drives, mobile hard disks, magnetic disks, or optical discs.

[0196] The present application also discloses a computer-readable storage medium storing instructions. When executed by one or more processors 301, it causes the electronic device to execute the method as described in one or more of the above embodiments.

[0197] The above are only exemplary embodiments of the present disclosure and should not be used to limit the scope of the present disclosure. That is, any equivalent changes and modifications made in accordance with the teachings of the present disclosure still fall within the scope covered by the present disclosure. After considering the specification and the disclosure of the practical truth, those skilled in the art will easily think of other implementation manners of the present disclosure. The present application aims to cover any variations, uses, or adaptation changes of the present disclosure, and these variations, uses, or adaptation changes follow the general principles of the present disclosure and include the well-known common knowledge or conventional technical means in the technical field not recorded in the present disclosure. The specification and the embodiments are only regarded as exemplary, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A method for collaborative lifting of a large-load unmanned aerial vehicle, characterized in that The method is applied to a virtual leading drone in a drone swarm, and the method includes: Obtaining the node positions and load distribution data of each edge node, where the virtual leading drone is the central node of the drone swarm, and the drones other than the virtual leading drone in the drone swarm are edge nodes; Calculating the total load data of the loads hoisted by the multiple edge nodes based on the load distribution data; Determining the overall motion trajectory of the drone swarm based on the total load data and the node positions, and sending the overall motion trajectory to the edge nodes, so that the edge nodes can find the required reference pulling force value and node motion trajectory based on the ant colony algorithm; During flight, if it is determined that the sling of the first drone among multiple drones breaks, synchronously distribute the pulling force to each second drone according to a predetermined load redistribution scheme, where the second drone is the drone other than the first drone in the drone swarm; Setting the first drone as the proxy central node and receiving the updated motion trajectory of the drone swarm from the proxy central node.

2. The method for collaborative hoisting of a large-load unmanned aerial vehicle according to claim 1, wherein The calculating the total load data of the loads hoisted by the multiple edge nodes based on the load distribution data specifically includes: Obtaining the load distribution data calculated by each edge node, where the edge node obtains the sling pulling force, its own vertical acceleration, and pitch and roll angles in real time through the mounted tension sensor, accelerometer, and attitude sensor, calculates the vertical component force borne by itself, and then corrects the inertial change of the suspension point based on the inertial acceleration measured by itself, calculates the effective gravitational acceleration correction factor under unit pulling force, and obtains the corrected load distribution data; Calculating the position of the center of gravity of the goods based on all the load distribution data as weights and the multiple node positions; Determining the pulling force action vector of the edge node according to the direction vector formed between the node position and the position of the center of gravity of the goods; Projecting the pulling force action vector in the direction of gravity to obtain the action component of each load distribution data in the vertical direction; Performing inertial correction on the action component according to the acceleration data of the drone swarm to obtain the corrected action component; Performing weighted fusion on the corrected action component according to the relative distance between the node position and the position of the center of gravity of the goods to obtain the total load data.

3. A method for collaborative lifting of a large-load unmanned aerial vehicle according to claim 1, characterized in that, The determining the overall motion trajectory of the drone swarm based on the total load data and the node positions specifically includes: Constructing a suspension point spatial distribution model of the drone swarm by combining the total load data and the node positions; Based on the suspension point spatial distribution model, and combining the change rate and instantaneous acceleration data of the position of the center of gravity of the goods, establishing a state vector of the center of mass of the goods, and then predicting the motion trend of the center of mass of the goods; On the basis of the motion trend of the center of mass of the goods, introducing mass distribution constraints, tension balance constraints, and attitude consistency constraints, constructing a load-node structure coupling model, so as to establish a dynamic mapping relationship between the motion trend of the center of mass of the goods and the flight path of the drone swarm; Introduce the multi-body system dynamics modeling method, regard the goods as a concentrated mass body, connect each of the edge nodes through flexible cables to form a spatial tension network structure, and establish the dynamic control equation of the entire UAV cluster based on the tensile force balance constraint and the angular momentum conservation constraint; According to the task flight path input and the movement trend of the center of mass of the goods, use the trajectory optimal control algorithm to solve the dynamic control equation to form the expected three-dimensional path of the center of mass of the goods; Based on the load-node structure coupling model, map the expected three-dimensional path to the flight path of the UAV cluster to obtain the overall movement trajectory.

4. A method for collaborative lifting of a large payload drone according to claim 1, characterized in that, Send the overall movement trajectory to the edge nodes, so that the edge nodes, based on the ant colony algorithm, search for the required reference tensile force value and the node movement trajectory, specifically including: Obtain the reference tensile force value and the node movement trajectory calculated by the edge nodes based on the overall movement trajectory. Among them, the edge nodes construct a local path search graph, and the path search graph consists of the current position of the node, the adjacent path state, the dynamic attitude limit and the cable geometric boundary, forming a feasible solution graph of the path space. Based on the path search graph, construct a heuristic search model, use the current expected path point in the overall movement trajectory as the target guidance, and control the transition probability through the pheromone intensity and the heuristic function; in each iteration, obtain the satisfaction degree of the path tension combination solution generated by the edge node evaluation ant colony for the system's tension balance degree and attitude consistency index, and define the objective function; after multiple rounds of iteration, the edge nodes select the node movement trajectory and the reference tensile force value corresponding to the optimal ant colony path as the execution command for the current control period, and send the node movement trajectory and the reference tensile force value to the central node.

5. A method for collaborative lifting of a large-load unmanned aerial vehicle according to claim 1, characterized in that During the flight, if it is determined that the first UAV among multiple UAVs has a suspension rope break, before synchronously distributing the tensile force to each second UAV according to the predetermined load redistribution scheme, the method further includes: Construct a suspension point spatial distribution model of the UAV cluster by combining the total load data and the node positions; Construct a virtual failure model for each edge node in the case of a suspension rope break. Without changing the total mass of the goods, respectively remove the load distribution data corresponding to each failed node among multiple edge nodes, and solve a new set of tension distributions for the remaining nodes, so as to minimize the perturbation cost function compared with the suspension point spatial distribution model while satisfying the conservation of the resultant tension force and the angular momentum constraint; Use a quadratic programming solver to perform offline enumeration calculation on the perturbation cost function, and generate an optimal redistribution solution corresponding to the suspension rope break for each failed node. Each optimal redistribution solution corresponds to a redistribution configuration for a failure case; Combine the optimal redistribution solutions corresponding to all the failed nodes to obtain the load redistribution scheme.

6. The method for collaborative lifting of a large-load unmanned aerial vehicle according to claim 5, wherein During the flight, if it is determined that the first UAV among multiple UAVs has a suspension rope break, synchronously distribute the tensile force to each second UAV according to the predetermined load redistribution scheme, specifically including: Determine the failed node corresponding to the first UAV; Invoke the optimal reallocation solution corresponding to the failed node corresponding to the first drone; Determine the remaining nodes corresponding to each of the second drones in the optimal reallocation solution, so as to determine the updated tension value and the unit vector of the cable direction that each of the second drones should bear; Broadcast the updated tension value and the unit vector of the cable direction to each of the second drones respectively.

7. A method for collaborative lifting of a large-load unmanned aerial vehicle according to claim 1, characterized in that, Before obtaining the node position and load distribution data of each edge node, the method further includes: Obtain the real-time state parameters broadcast by each drone in the drone cluster during the takeoff preparation phase, where the real-time state parameters include position information, remaining computing resources, communication signal quality, historical trajectory stability index, and node reliability level; Calculate the centrality score of the virtual pilot drone according to the real-time state parameters, and broadcast the centrality score of the virtual pilot drone; After completing the omnidirectional score synchronization and each drone independently compares the magnitudes of the centrality scores of multiple drones, if it is determined that the centrality score of the virtual pilot drone is the highest, then determine the virtual pilot drone as the central node.

8. A large-load unmanned aerial vehicle collaborative lifting device, characterized in that, The device is used to execute a large-load drone collaborative hoisting method according to any one of claims 1-7. The device is arranged on a virtual pilot drone in the drone cluster. The device includes an acquisition module (201), a processing module (202), and an output module (203), where: The acquisition module (201) is used to obtain the node position and load distribution data of each edge node, where the virtual pilot drone is the central node of the drone cluster, and the drones in the drone cluster except the virtual pilot drone are edge nodes; The processing module (202) is used to calculate the total load data of the loads hoisted by multiple edge nodes based on the load distribution data; The processing module (202) is used to determine the overall movement trajectory of the drone cluster based on the total load data and the node positions, and send the overall movement trajectory to the edge nodes, so that the edge nodes can find the required reference tension value and node movement trajectory based on the ant colony algorithm; The output module (203) is used to, during flight, if it is determined that the suspension rope of the first drone among multiple drones breaks, synchronize the tension distribution to each second drone according to a predetermined load redistribution scheme, where the second drones are the drones in the drone cluster except the first drone; The output module (203) is used to set the first drone as the proxy central node and receive the updated movement trajectory of the drone cluster from the proxy central node.

9. An electronic device, characterized in that, It includes a processor (301), a communication bus (302), a user interface (303), a network interface (304), and a memory (305). The memory (305) is used to store instructions. Both the user interface (303) and the network interface (304) are used to communicate with other devices. The communication bus (302) is used to realize the connection and communication between components within the electronic device. The processor (301) is used to execute the instructions stored in the memory (305) so that the electronic device executes the method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions which, when executed, perform the method according to any one of claims 1-7.

Images