Methods and operating platforms for coordinated control of multi-mission mounted equipment on unmanned aerial vehicles (UAVs)

Abstract

This invention relates to the field of UAV flight control technology, and more particularly to a method and platform for the coordinated control of multi-mission payload equipment on UAVs. It includes: constructing a task chain with time weights; dynamically monitoring the remaining time weights and correcting the consumption rate, triggering a window adjustment period when it decreases to a preset threshold; determining the task chain transition state within the window period through effect verification logic; issuing a feedforward blanking signal based on the transition state to gradually blank the current dynamic compensation amount, while simultaneously preloading feedforward parameters to the flight controller based on the characteristics of the next mission payload and issuing a payload pre-activation command to achieve online superposition and fusion of control quantities; and adaptively adjusting the task chain execution boundary in case of abnormal states. This invention achieves smooth connection between the flight controller state and multiple payloads during multi-mission switching, significantly improving the control stability and autonomy of continuous UAV operation.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) flight control technology, and in particular to a method and platform for the coordinated control of multi-task mounted equipment on UAVs. Background Technology

[0002] With the dense development of high-rise buildings in cities, grassroots social and political security and ideological security are facing new and sudden risks. A small number of criminals are taking advantage of the concealed locations of high-rise buildings to maliciously project illegal anti-propaganda slogans onto the walls or windows of opposite buildings at night using high-power lasers, or to hang illegal anti-propaganda banners from high altitudes. Such behaviors are characterized by high locations, concealed operations, late detection, and rapid spread. Traditional prevention and control methods mainly rely on manual ground patrols and subsequent building-based cleanup. This approach is not only limited by blind spots on the ground, making it difficult to detect long-distance cross-building projections with the naked eye in a timely manner, but also has a very short window for intervention. Furthermore, manual building-based cleanup poses significant safety hazards due to the high-altitude operations.

[0003] However, existing control technologies have the following shortcomings when using industrial drones to continuously perform high-altitude special operations such as laser blocking, source interception, and banner coverage. First, timing control lacks an adaptive mechanism, failing to dynamically adjust the time weights of each task node by combining load start / stop delays and environmental wind field interference, resulting in insufficient timing matching. Second, dynamic shocks occur during task switching; the mass reduction and aerodynamic shape changes caused by load actuation lead to combined disturbances in the drone's center of gravity and aerodynamic characteristics. Traditional feedback regulation struggles to eliminate attitude deviations during the transition period and lacks a smooth transition mechanism between old and new control variables. Third, there is a lack of effect closed-loop and reconfiguration capabilities, making it impossible to quantitatively evaluate the on-site operation results online. When operations fail to meet standards or system anomalies occur, it cannot adaptively extend time weights or adjust task chain execution boundaries, limiting the stable control capability for continuous operations. Summary of the Invention

[0004] To overcome the above deficiencies, this invention provides a method and platform for coordinated control of multi-task payload equipment on unmanned aerial vehicles (UAVs), aiming to achieve smooth connection between flight control mode and multiple payloads during multi-task switching, and improve the control stability and autonomy of UAVs in continuous operation.

[0005] In a first aspect, the present invention provides the following technical solution: a method for coordinated control of multi-task mounted operation equipment on unmanned aerial vehicles (UAVs), comprising:

[0006] S1. Obtain the perception data of the work scenario, construct the work task chain with time weights, and extract the temporal features and control parameters of the current task node.

[0007] S2. Control the multiple types of mounted equipment to execute the current task along the task chain, and trigger the window adjustment period when the remaining time weight of the current task node decreases to a preset threshold.

[0008] S3. Collect operation effect data during the window adjustment period, and determine the current task chain flow status through effect verification logic;

[0009] S4. Based on the aforementioned flow state, a feedforward blanking signal is sent to the flight control system to gradually blank the dynamic compensation amount corresponding to the current task, so as to smoothly transition the physical state.

[0010] S5. Based on the physical characteristics of the load of the next task node to be executed, preload aerodynamic and center of mass change feedforward parameters into the flight control system, and generate a pre-activation command to be sent to the corresponding mounted equipment.

[0011] S6. Monitor the actual feedback data during task execution and switching. When an abnormal state occurs, adaptively adjust the execution boundary of the task chain.

[0012] Preferably, in S1, the task chain is implemented using a directed acyclic graph, and the step of constructing the task chain with time weights includes:

[0013] The task scenario is analyzed to deconstruct multiple task subtasks and the topological dependencies between them.

[0014] The topological nodes and directed edges of the directed acyclic graph are generated based on the topological dependency mapping.

[0015] Based on the time constraints of the job subtask, configure a first time weight representing the expected execution time for the topology node;

[0016] Based on the UAV's maneuvering restrictions or payload switching delay, a second time weight representing the flow delay constraint is configured for the directed edge.

[0017] Preferably, in S1, the step of extracting the temporal features and control parameters of the current task node includes:

[0018] The real-time execution progress of the job scenario is analyzed, and the remaining executable window of the current task node is calculated based on the time weight to generate the temporal features that characterize the node flow boundary.

[0019] The operating status parameters and physical change data of the various types of mounted equipment are obtained, and the dynamic disturbance parameters of the whole system caused by load changes are simulated online through the airborne dynamics solution model.

[0020] Preferably, in S2, the step of triggering the window adjustment period includes:

[0021] The minimum convergence time is calculated as a preset threshold based on the load start-up and shutdown delay of the next task node to be executed and the current environmental interference.

[0022] Monitor the dynamic completion rate of the current task and adjust the consumption rate of the remaining time weight of the current task node based on the deviation between the dynamic completion rate and the time already consumed.

[0023] The remaining time weight is compared with the preset threshold in real time. When the remaining time weight decreases to the preset threshold, the window adjustment period is triggered.

[0024] Preferably, in S3, the step of determining the current task chain's flow state through effect verification logic includes:

[0025] Acquire real-time sensing data of the target area and extract residual attribute information of the target to be treated;

[0026] After spatial registration, the feature data before and after the operation are compared, and a quantitative evaluation index characterizing the degree of elimination of the target to be treated is calculated.

[0027] The quantitative evaluation index is compared with the preset safety threshold. If the threshold is met, a forward flow state variable is output to advance the task chain; otherwise, a reverse flow state variable is output to trigger node renewal.

[0028] Preferably, in S4, the step of gradually eliminating the dynamic compensation amount corresponding to the current task includes:

[0029] Based on the remaining duration of the window adjustment period, a time-varying blanking coefficient that monotonically decreases over time is constructed.

[0030] The time-varying blanking coefficient is used to attenuate the dynamic compensation amount of the current task and generate a real-time attenuation control amount.

[0031] The real-time attenuation control quantity is dynamically injected into the attitude control loop of the flight control system, and the dynamic compensation quantity is controlled to smoothly converge with the preload feedforward parameters of the next task to be executed at the end of the window adjustment period.

[0032] Preferably, in S5, the step of preloading aerodynamic and center-of-gravity change feedforward parameters into the flight control system includes:

[0033] Based on the physical characteristics of the load of the next task node to be executed, calculate the predicted value of the centroid offset and the predicted value of the aerodynamic parameter change.

[0034] Based on the remaining duration of the window adjustment period, the predicted quantity is subjected to temporal smoothing mapping to generate a gradually increasing feedforward compensation sequence.

[0035] During the window adjustment period, the feedforward compensation sequence is gradually injected, so that it is superimposed and fused with the real-time attenuation control quantity online.

[0036] Preferably, in S6, the step of adaptively adjusting the execution boundary of the job task chain includes:

[0037] When the abnormal state is that the task fails to meet the target, the time weight of the current task node is extended.

[0038] When the abnormal state is a system logic failure or a sudden environmental change, the topology reconstruction of the job task chain is performed.

[0039] Secondly, the present invention also provides a collaborative operation platform for multi-task mounted operation equipment of unmanned aerial vehicles (UAVs), used to execute the collaborative control method described above, wherein the operation platform includes:

[0040] The task chain configuration module is used to acquire the perception data of the work scenario, construct the task chain with time weights, and extract the temporal features and control parameters of the current task node.

[0041] The timing coordination control module is used to control multiple types of mounted equipment to execute the current task along the task chain, and to trigger a window adjustment period when the remaining time weight of the current task node decreases to a preset threshold.

[0042] The effect verification module is used to collect operation effect data during the window adjustment period and determine the current task chain flow status through effect verification logic.

[0043] The feedforward blanking module is used to send a feedforward blanking signal to the flight control system according to the flow state, and to gradually blank the dynamic compensation amount corresponding to the current task so as to smoothly transition the physical state.

[0044] The load preloading and activation module is used to preload aerodynamic and center of mass change feedforward parameters into the flight control system based on the load physical characteristics of the next task node to be executed, and generate a pre-activation command to be sent to the corresponding mounted equipment.

[0045] The adaptive reconfiguration module is used to monitor the actual feedback data during task execution and switching. When an abnormal state occurs, it adaptively adjusts the execution boundary of the job task chain.

[0046] The present invention has the following beneficial effects:

[0047] 1. In this invention, a smooth transition of multiple load control quantities is achieved through online superposition and fusion of the blanking coefficient and the preload feedforward parameter. This effectively eliminates dynamic disturbances caused by load switching and centroid abrupt changes, avoids airframe excitation caused by sudden changes in control signals, and ensures the attitude stability of the UAV during mission transitions.

[0048] 2. In this invention, a flexible spatiotemporal safety boundary is constructed based on a task chain with time weights and an adaptive reconstruction mechanism. When faced with anomalies such as unmet task requirements or sudden environmental changes, the system can dynamically perform time weight extension or topology reconstruction, avoiding task chain timing collapse and significantly improving the system's emergency fault tolerance and overall survivability. Attached Figure Description

[0049] Figure 1 The flowchart below shows the collaborative control method for multi-task mounted equipment of unmanned aerial vehicles (UAVs) proposed in this invention.

[0050] Figure 2 This is an architecture diagram of the collaborative operation platform for multi-task mounted equipment of unmanned aerial vehicles proposed in this invention. Detailed Implementation

[0051] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] Example 1

[0053] In the first embodiment of the present invention, the present invention provides a method and platform for coordinated control of multi-task mounted operation equipment for unmanned aerial vehicles (UAVs), such as... Figure 1 As shown, it includes the following steps:

[0054] S1. Obtain the perception data of the work scenario, construct the work task chain with time weights, and extract the temporal features and control parameters of the current task node.

[0055] Preferably, in S1, the task chain is implemented using a directed acyclic graph, and the step of constructing the task chain with time weights includes:

[0056] The task scenario is analyzed to deconstruct multiple task subtasks and the topological dependencies between them.

[0057] The topological nodes and directed edges of the directed acyclic graph are generated based on the topological dependency mapping.

[0058] Based on the time constraints of the job subtask, configure a first time weight representing the expected execution time for the topology node;

[0059] Based on the UAV's maneuvering restrictions or payload switching delay, a second time weight representing the flow delay constraint is configured for the directed edge.

[0060] Specifically, the system first obtains the current input task scenario description file, which includes the overall task objective of the task, the traversable airspace boundary, the three-dimensional coordinates of obstacles, and the distribution of target task points. The system then uses a preset semantic parser to deconstruct the task scenario, extracting multiple task subtasks that have independent execution characteristics in time or space, and simultaneously identifying the topological dependencies between the various task subtasks.

[0061] Based on the topological dependencies derived above, the topological nodes and directed edges of the directed acyclic graph are mapped and generated. The mathematical topological structure of the directed acyclic graph is as follows:

[0062] ;

[0063] in, This represents the constructed directed acyclic graph; This represents the set of topological nodes in a directed acyclic graph. Represents a set The first in One topology node; Represents the set of directed edges; Indicates the topology nodes Point to topology node The directed edge.

[0064] Based on the time constraints of each job subtask, the system assigns a first time weight representing the expected execution time to each topological node in the directed acyclic graph. The calculation process for this weight is as follows:

[0065] ;

[0066] in, This represents the node time weight mapping function; Indicates the first The nominal execution time of each subtask under the baseline operating conditions; This represents the safety time redundancy reserved to cope with sudden environmental changes or operational delays. The physical meaning of this formula lies in quantifying the net time overhead required for a single subtask to perform its operation at a spatial target point. The first time weight not only includes the physical time required for the operational subject, such as UAVs, to perform data acquisition or positioning detection, but also incorporates a safety time boundary to ensure that the operational behavior of a single node has a defined time span.

[0067] Based on the maneuvering restrictions or payload switching delays of the UAV, a second time weight representing the flow delay constraint is assigned to each directed edge in the directed acyclic graph. The calculation process of this weight is as follows:

[0068] ;

[0069] in, This represents the time weight mapping function for directed edges; This represents the minimum maneuver time required for a UAV to travel from the starting point to the ending point of the directed edge in the track space; This represents the time delay incurred by the UAV's payload during hardware mode switching or preheating and stabilization before entering the next node. The physical meaning of this formula lies in quantifying the unavoidable physical delay that occurs when the system state transitions between different subtasks. The second time weight directly translates the UAV's underlying kinematic constraints and the payload controller's hardware delay into edge weights of a directed acyclic graph, enabling the upper-layer task chain to possess the underlying physical executability.

[0070] Preferably, in S1, the step of extracting the temporal features and control parameters of the current task node includes:

[0071] The real-time execution progress of the job scenario is analyzed, and the remaining executable window of the current task node is calculated based on the time weight to generate the temporal features that characterize the node flow boundary.

[0072] The operating status parameters and physical change data of the various types of mounted equipment are obtained, and the dynamic disturbance parameters of the whole system caused by load changes are simulated online through the airborne dynamics solution model.

[0073] Specifically, the system obtains the real-time cumulative execution time of the current task node by reading the onboard timer or receiving an external clock synchronization signal. Based on pre-configured time weights, the system calculates the remaining executable window of the current task node online to generate temporal characteristics representing the node's transition boundary. The calculation process of the remaining executable window is as follows:

[0074] ;

[0075] in, Indicates the current task node The remaining executable windows; Indicates assignment to the current task node Total time budget value; Indicates from the current task node The real-time execution progress time consumed from the start of execution to the current sampling time; Indicates origin from the current task node Point to the next target node directed edges The second time weight. The physical meaning of this formula lies in quantifying the net time limit that the task entity is allowed to remain within the current task node without violating the overall timing boundary. By subtracting the consumed time and downstream flow delay from the total budget, it provides dynamic hard time boundary characteristics for real-time scheduling decisions.

[0076] The system acquires the operational status parameters and physical change data of various types of mounted equipment. The operational status parameters are collected in real time through encoders or displacement sensors installed on the load mechanism, including the extension angle of the mounted equipment's robotic arm or the telescopic displacement of the equipment's slide rails. The physical change data includes the mass loss and sudden changes in rotational inertia caused by the mounted equipment's release, consumption, or structural reconfiguration. The system inputs this data into a preset airborne dynamics calculation model to online deduce the overall system dynamics disturbance parameters caused by load changes, using this as the basis for online correction of control parameters. The calculation formulas for the overall system dynamics disturbance parameters are as follows:

[0077] ;

[0078] in, This represents the dynamic disturbance parameter of the whole system, which is specifically manifested as the additional disturbance torque vector caused by load changes; This represents the change in the quality of the mounted equipment determined by the physical change data; Represents the gravitational acceleration constant; This represents the relative position vector of the mounted equipment's center of mass with respect to the current center of mass of the entire system. This represents the change matrix of the moment of inertia of the mounted equipment relative to the rotation axis of the whole machine, determined by the actuation state parameters. This represents the current angular acceleration vector of the drone.

[0079] S2. Control the multiple types of mounted equipment to execute the current task along the task chain, and trigger the window adjustment period when the remaining time weight of the current task node decreases to a preset threshold.

[0080] Preferably, in S2, the step of triggering the window adjustment period includes:

[0081] The minimum convergence time is calculated as a preset threshold based on the load start-up and shutdown delay of the next task node to be executed and the current environmental interference.

[0082] Monitor the dynamic completion rate of the current task and adjust the consumption rate of the remaining time weight of the current task node based on the deviation between the dynamic completion rate and the time already consumed.

[0083] The remaining time weight is compared with the preset threshold in real time. When the remaining time weight decreases to the preset threshold, the window adjustment period is triggered.

[0084] Specifically, in order to ensure that the system has sufficient time elasticity to perform attitude recovery and load configuration when the upper time chain is squeezed or the lower physical environment is disturbed, the system evaluates and triggers a window adjustment period online.

[0085] The system calculates the minimum convergence time as a preset threshold based on the load start / stop delay of the next task node to be executed and the current environmental interference. The calculation formula for the preset threshold is as follows:

[0086] ;

[0087] in, This represents the preset threshold, which specifically controls the minimum convergence time required for the algorithm to converge. Indicates the load start / stop delay for the next task node to be executed; This represents the preset interference conversion coefficient, which is specifically set to 0.6 seconds per meter per second in this embodiment; This indicates the current level of environmental disturbance, specifically manifested as the real-time external wind speed value collected by airborne environmental sensors.

[0088] The system monitors the dynamic completion rate of the current task and, based on the deviation between the dynamic completion rate and the elapsed time, adjusts the consumption rate of the remaining time weight of the current task node. The formula for calculating the adjusted consumption rate is as follows:

[0089] ;

[0090] in, This indicates the rate at which the remaining time weight of the current task node is adjusted and consumed. This represents the preset basic consumption rate constant, which is specifically set to 1.0 in this embodiment; This represents the preset schedule deviation adjustment coefficient, which is specifically set to 0.4 in this embodiment; This indicates the elapsed time for the current task node; This represents the nominal total budget time for the current task node; This represents the dynamic completion rate of the current task, specifically the ratio of the number of completed inspection trackpoints read via the airborne bus to the total number of trackpoints at the current node. The physical significance of this formula lies in introducing a feedback adjustment mechanism to offset the slippage error between upper-level planning and actual physical execution. When the actual task progress lags behind the time consumption progress, this formula accelerates the deduction of virtual remaining time by increasing the consumption rate. This creates a sense of urgency in the time dimension, even with spatial lag, forcing the system to enter an adjustment and preparation state earlier.

[0091] The system compares the remaining time weight with the preset threshold in real time. Within each control cycle, the system continuously updates the real-time remaining time weight of the current task node through integral decrement. When the remaining time weight decreases to the preset threshold... At that time, the system immediately issues a trigger command to control the system to enter the window adjustment period.

[0092] S3. Collect operation effect data during the window adjustment period, and determine the current task chain flow status through effect verification logic;

[0093] Preferably, in S3, the step of determining the current task chain's flow state through effect verification logic includes:

[0094] Acquire real-time sensing data of the target area and extract residual attribute information of the target to be treated;

[0095] After spatial registration, the feature data before and after the operation are compared, and a quantitative evaluation index characterizing the degree of elimination of the target to be treated is calculated.

[0096] The quantitative evaluation index is compared with the preset safety threshold. If the threshold is met, a forward flow state variable is output to advance the task chain; otherwise, a reverse flow state variable is output to trigger node renewal.

[0097] Specifically, in order to ensure that each task node achieves the expected operational effect before the workflow is completed, and to avoid safety hazards caused by blindly advancing the task chain, the system dynamically performs online effect verification and workflow control.

[0098] The system acquires real-time sensing data of the target area and extracts residual attribute information of the target to be treated. During this process, airborne sensors scan the target physical area after the operation has been completed, acquiring real-time sensing data containing high-resolution images or point cloud data. The system uses a preset feature extraction algorithm to separate the geometric contour, reflectivity, or spectral features of the target in its current state, forming digitally represented residual attribute information.

[0099] The system performs three-dimensional spatial registration between the baseline feature data stored before the operation and the feature data acquired in real time after the operation, within a unified spatial coordinate system. After registration, the system compares the feature data before and after the operation and calculates a quantitative evaluation index characterizing the degree of elimination of the target to be treated. The calculation formula for the quantitative evaluation index is as follows:

[0100] ;

[0101] in, This represents a quantitative assessment indicator, specifically the elimination rate of the target to be addressed; This represents the area of ​​the target feature region to be processed after spatial registration before the operation; This indicates the area of ​​the residual characteristic region of the target to be treated after spatial registration following the operation.

[0102] The system compares the quantitative evaluation indicators with preset safety thresholds and establishes different flow branches based on the comparison results. If the quantitative evaluation indicators reach or exceed the preset safety thresholds, a forward flow state variable is output to advance the task chain; if the quantitative evaluation indicators do not reach the preset safety thresholds, a reverse flow state variable is output to trigger node renewal. The determination logic formula for the state flow is as follows:

[0103] ;

[0104] in, Represents the output flow state variables; This represents the preset forward-flowing state variable identifier code, used to drive the topology pointer of the directed acyclic graph to advance backward; This indicates the preset reverse flow state variable identifier code, which is used to trigger the timing re-initialization and time weight renewal of the current node; Indicates quantitative evaluation indicators; This represents a preset safety threshold, specifically set to 0.85 in this embodiment. The physical meaning of this formula lies in constructing a discrete control switch for the task chain flow. By logically mapping continuous performance evaluation metrics to rigid physical safety boundaries, it determines whether to continue executing downstream subtasks in the directed acyclic graph or to forcibly halt at the current task node and perform secondary tasks.

[0105] S4. Based on the aforementioned flow state, a feedforward blanking signal is sent to the flight control system to gradually blank the dynamic compensation amount corresponding to the current task, so as to smoothly transition the physical state.

[0106] Preferably, in S4, the step of gradually eliminating the dynamic compensation amount corresponding to the current task includes:

[0107] Based on the remaining duration of the window adjustment period, a time-varying blanking coefficient that monotonically decreases over time is constructed.

[0108] The time-varying blanking coefficient is used to attenuate the dynamic compensation amount of the current task and generate a real-time attenuation control amount.

[0109] The real-time attenuation control quantity is dynamically injected into the attitude control loop of the flight control system, and the dynamic compensation quantity is controlled to smoothly converge with the preload feedforward parameters of the next task to be executed at the end of the window adjustment period.

[0110] Specifically, in order to eliminate the disturbances caused to the machine body by sudden changes in the underlying control signals when different tasks are switched, the system performs alternating fading of control parameters and smooth feedforward connection during the window adjustment period.

[0111] The system constructs a time-varying blanking coefficient that monotonically decreases over time based on the remaining duration of the window adjustment period. After the window adjustment period begins, the onboard timer tracks the current adjustment period's progress in real time and calculates the remaining duration by combining it with the preset total adjustment period time overhead, thereby mapping and generating the time-varying blanking coefficient. The calculation formula for the time-varying blanking coefficient is as follows:

[0112] ;

[0113] in, Indicates the current moment The time-varying blanking coefficient; This indicates that the window adjustment period is at the current time. The remaining time; This indicates the preset total duration of the window adjustment period, which is specifically set to 3.0 seconds in this embodiment.

[0114] The system uses the time-varying blanking coefficient to attenuate the dynamic compensation amount of the current task, generating a real-time attenuation control quantity. It also periodically acquires the time-varying blanking coefficient calculated in the current step and uses it as a gain factor to perform a scalar multiplication with the current dynamic compensation amount. The calculation formula for the real-time attenuation control quantity is as follows:

[0115] ;

[0116] in, This represents the generated real-time decay control vector; Indicates the current moment The time-varying blanking coefficient; This represents the dynamic compensation vector generated by the current task and that needs to be hidden.

[0117] The system dynamically injects the real-time attenuation control quantity into the attitude control loop of the flight control system, ensuring smooth convergence of the dynamic compensation quantity with the preloaded feedforward parameters of the next task at the end of the window adjustment period. During this process, the loop mixer receives the real-time attenuation control quantity of the current task and the feedforward parameters of the next task. After alternating and overlapping calculations of the execution control quantities, the mixed control quantity is applied to the attitude servo actuator in real time. The formula for injecting the mixed control quantity into the attitude control loop is as follows:

[0118] ;

[0119] in, This represents the total control vector that is ultimately dynamically injected into the attitude control loop of the flight control system. This represents the current real-time decay control vector; Indicates the current moment The time-varying blanking coefficient; This represents the preloaded feedforward parameter vector for the next task to be executed. The physical meaning of this formula lies in constructing a linear interpolation transition space between two different task dynamic modes. During the window adjustment period, as the blanking coefficient approaches 0 from 1, the control contribution of the current task gradually decays linearly to zero, while the feedforward control component of the next task to be executed is simultaneously and smoothly filled from zero to its full amplitude. When the window adjustment period ends precisely at the boundary point, the total control quantity coincides with the preloaded feedforward parameters of the next task, achieving a bounded and seamless transfer of overall control during discrete task switching, ensuring the attitude stability of the UAV during the flight task transition period.

[0120] S5. Based on the physical characteristics of the load of the next task node to be executed, preload aerodynamic and center of mass change feedforward parameters into the flight control system, and generate a pre-activation command to be sent to the corresponding mounted equipment.

[0121] Preferably, in S5, the step of preloading aerodynamic and center-of-gravity change feedforward parameters into the flight control system includes:

[0122] Based on the physical characteristics of the load of the next task node to be executed, calculate the predicted value of the centroid offset and the predicted value of the aerodynamic parameter change.

[0123] Based on the remaining duration of the window adjustment period, the predicted quantity is subjected to temporal smoothing mapping to generate a gradually increasing feedforward compensation sequence.

[0124] During the window adjustment period, the feedforward compensation sequence is gradually injected, so that it is superimposed and fused with the real-time attenuation control quantity online.

[0125] Specifically, in order to enable the UAV to adapt in advance to the changes in physical characteristics brought about by the upcoming change in mission payload, the system calculates and preloads future feedforward control quantities in parallel during the window adjustment period.

[0126] The system calculates the predicted center of mass shift and aerodynamic parameter changes based on the load physical characteristics of the next task node to be executed. During this process, the system reads the structural dimensions, nominal mass, and installation axis offset data of the next task's mounted equipment stored in the configuration table. Through a preset geometric-mechanical mapping mechanism, it calculates the expected change in the overall center of mass position caused by the new load, i.e., the predicted center of mass shift. Simultaneously, it combines this with the current flow field prediction benchmark to calculate the correction values ​​for the frontal area and drag coefficient resulting from the change in the new load's shape, i.e., the predicted aerodynamic parameter changes. The calculation formulas for the benchmark feedforward parameters are as follows:

[0127] ;

[0128] in, This represents the reference feedforward parameter vector for the next task node to be executed; This represents the preset mass-gravity conversion coefficient matrix; This represents the calculated centroid offset prediction vector; This represents the preset aerodynamic torque conversion coefficient matrix; This represents the vector of predicted changes in aerodynamic parameters obtained from the calculation.

[0129] The system acquires the current window adjustment period time progress in real time and dynamically spreads the calculated static baseline feedforward parameters onto the time axis using mapping weights complementary to the blanking coefficients. The formula for calculating the gradually increasing feedforward compensation sequence is as follows:

[0130] ;

[0131] in, Indicates the current moment The real-time control component vector of the feedforward compensation sequence; This represents the time-varying blanking coefficient that is monotonically decreasing with time, calculated in step S4. This represents the feedforward parameter vector.

[0132] During the window adjustment period, the system gradually injects feedforward compensation sequences, which are then superimposed and fused online with the real-time attenuation control quantity. Within each control sampling cycle of the window adjustment period, the airborne bus driver sends the dynamic feedforward control component calculated in the current step to the core calculation chip of the flight control system in real time. This component is then vector-added at the same control level with the real-time attenuation control quantity from the current task in the attitude control loop, performing the same-dimensional addition. The feedforward parameters are fused online using the following formula:

[0133] ;

[0134] in, This represents the total fused control vector that is finally output to the servo control loop after online overlay and fusion; This represents the real-time attenuation control vector generated in step S4; Indicates the current moment The real-time control component vector of the feedforward compensation sequence. The physical meaning of this formula is to achieve a smooth physical transition between the old and new task dynamics compensation mechanisms in the total control output channel. By linearly superimposing a monotonically decreasing attenuation control line and a monotonically increasing feedforward compensation line within the same time window, the total control torque of the system maintains dynamic energy conservation and stable change at the moment of load switching, thereby completely eliminating attitude fluctuations caused by changes in the morphology of the mounted equipment.

[0135] S6. Monitor the actual feedback data during task execution and switching. When an abnormal state occurs, adaptively adjust the execution boundary of the task chain.

[0136] Preferably, in S6, the step of adaptively adjusting the execution boundary of the job task chain includes:

[0137] When the abnormal state is that the task fails to meet the target, the time weight of the current task node is extended.

[0138] When the abnormal state is a system logic failure or a sudden environmental change, the topology reconstruction of the job task chain is performed.

[0139] Specifically, when the abnormal state is that the task has not met the target, i.e., the flow status variable output in step S3 is a reverse flow status variable, the system extends the time weight of the current task node. During this process, the system dynamically injects incremental time resources to reinitialize the execution cycle of the current task node. The formula for calculating the new time weight after the extension is as follows:

[0140] ;

[0141] in, Indicates the current task node The new time weights after the extension; Indicates the current task node The initial time weights of the original configuration; This indicates the preset single extension step size, which is specifically set to 30 seconds in this embodiment.

[0142] When the abnormal state is a system logic failure or a sudden environmental change, such as when the onboard self-test detects hardware malfunction or environmental sensors detect wind speed exceeding physical safety boundaries, the system performs topology reconstruction of the task chain. During this process, the system performs discrete pruning and path realignment on the node and edge sets of the directed acyclic graph. The mapping formula for the topology reconstruction is as follows:

[0143]

[0144] in, This represents the newly reconstructed directed acyclic graph data block; This represents a predefined graph topology reconstruction operator, which is specifically a directed edge reconstruction procedure that removes unreachable nodes and reconnects the remaining nodes using Dijkstra's algorithm. This represents the original directed acyclic graph with time weights before reconstruction; This represents a set of faulty nodes or high-risk environment nodes that have been identified as unable to continue execution. When external physical constraints undergo irreversible and drastic changes, the system disconnects all incoming and outgoing edges associated with the damaged or dangerous nodes and recalculates a physically feasible optimal timing chain among the remaining safe topology nodes, thereby fundamentally ensuring the overall survivability of the UAV in extreme and sudden scenarios.

[0145] Example 2

[0146] This invention provides a collaborative operation platform for multi-task mounting equipment on unmanned aerial vehicles (UAVs), the structure of which is as follows: Figure 2 As shown. The operating platform includes:

[0147] The task chain configuration module is used to acquire the perception data of the work scenario, construct the task chain with time weights, and extract the temporal features and control parameters of the current task node.

[0148] The timing coordination control module is used to control multiple types of mounted equipment to execute the current task along the task chain, and to trigger a window adjustment period when the remaining time weight of the current task node decreases to a preset threshold.

[0149] The effect verification module is used to collect operation effect data during the window adjustment period and determine the current task chain flow status through effect verification logic.

[0150] The feedforward blanking module is used to send a feedforward blanking signal to the flight control system according to the flow state, and to gradually blank the dynamic compensation amount corresponding to the current task so as to smoothly transition the physical state.

[0151] The load preloading and activation module is used to preload aerodynamic and center of mass change feedforward parameters into the flight control system based on the load physical characteristics of the next task node to be executed, and generate a pre-activation command to be sent to the corresponding mounted equipment.

[0152] The adaptive reconfiguration module is used to monitor the actual feedback data during task execution and switching. When an abnormal state occurs, it adaptively adjusts the execution boundary of the job task chain.

[0153] The following example, using a real-world emergency response scenario involving illegal high-altitude banners, illustrates the application process of the UAV multi-task mounted equipment collaborative control method and collaborative operation platform of this invention.

[0154] A local police station received a report from residents that illegally spray-painted slogans had appeared on the exterior facade of a 28-story residential building in its jurisdiction. The slogans, located approximately 2.5 square meters on the east side of the building at a height of about 22 stories, contained illegal propaganda and required immediate action. The drone used in the operation was a six-rotor industrial-grade drone. The equipment mounted on the drone included: a 75,000-lumen high-intensity light source to interfere with laser-projection-based anti-propaganda activities; a high-pressure spray coating system equipped with a dedicated coating material storage tank and a controllable flow nozzle; and a high-definition visible light and infrared dual-light imaging system for collecting sensory data and verifying the effects before and after the operation.

[0155] After the drone arrives at the site, the task chain configuration module acquires the scene perception data and deconstructs the overall task into a directed acyclic graph containing three topological nodes and two directed edges: Node 1 is target localization and confirmation (activation of dual-light imaging equipment); Node 2 is spray coating treatment (switching to spray coating equipment); Node 3 is strong light shielding and final verification (activation of strong light equipment and camera recording). The first time weight of Node 1 is configured as 15 seconds; Node 2 is configured as 8 seconds based on an area of ​​2.5 square meters; Node 3 is configured as 9 seconds. The second time weight of the transition from Node 1 to Node 2 is 7.4 seconds (2.4s maneuver + 5s delay); the second time weight of the transition from Node 2 to Node 3 is 7.1 seconds. During the execution of Node 2, the module simulates the dynamic disturbances caused by spray coating in real time. The formula for the dynamic disturbance parameter simulation of the entire system is:

[0156]

[0157] Among them, the calculated Specifically, this manifests as a forward tilting moment of the center of mass caused by the spraying load (approximately 0.15 N·m) and a vibration disturbance moment of the spraying pump body (approximately 0.05 N·m).

[0158] The timing-coordinated control module controls the execution of node two along the task chain. The module monitors the remaining time weight of node two in real time (initial 8 seconds, baseline rate 1.0). When the spraying reaches the 4th second, the area covered exceeds the expected area. The module dynamically adjusts the consumption rate according to the deviation formula, which is as follows:

[0159] ;

[0160] At the same time, the preset threshold for solving node two is calculated:

[0161] ;

[0162] Among them, the load start / stop delay of the next node (node ​​three) The interference conversion coefficient is 2.5 seconds. The current upper-level sluice wind speed is 0.6. The speed is 1.66 m / s, and the preset threshold is calculated. = 3.5 seconds. When the remaining time weight of node two decreases to 3.5 seconds, the system triggers the window adjustment period.

[0163] After entering the window adjustment period of Node 2, the effect verification module collects operation effect data through the device's built-in close-range visual sensor and calculates quantitative evaluation indicators:

[0164] ;

[0165] in, The area of ​​the slogan before spraying. This represents the remaining visible area. The module injects the calculation results into the decision logic. Due to the actual solution coverage The value was 0.98, successfully exceeding the preset safety threshold. The system outputs the downstream state variable. This allows the task chain to advance towards node three.

[0166] Based on the current flow state, the feedforward blanking module performs gradual blanking on the dynamic compensation amount corresponding to the current spraying operation. The module calculates the time-varying blanking coefficient based on the remaining duration of the window adjustment period.

[0167] ;

[0168] Among them, the total duration of the preset adjustment period The time is 3.0 seconds. The module uses this coefficient to generate real-time decay control. The dynamic injection of this force into the attitude control loop allows the 0.15 N·m forward tilting torque compensation to linearly and smoothly return to zero within 3.0 seconds.

[0169] During the blanking period, the load preloading and activation module calculates the feedforward parameters based on the physical characteristics of the node's three loads (a combination of high-intensity light and imaging equipment, weighing 1.2 kg, with a mounting lever arm of 0.25 m):

[0170] ;

[0171] The centroid offset prediction is obtained by solving the problem. (0.02 meters along the X-axis, 0.008 meters along the Z-axis) and the reduction in aerodynamic drag. The module generates a feedforward compensation sequence. Gradually injected and superimposed and integrated with the real-time attenuation control quantity online. Simultaneously, a pre-activation command is sent to the high-intensity imaging device, enabling it to complete unlocking and self-testing within 3.0 seconds. At the end of the window adjustment period, the new payload feedforward takes over at full amplitude, and the physical state transitions smoothly.

[0172] Throughout the entire task execution and switching process, the adaptive reconstruction module continuously monitors the actual feedback data: if the effect verification module detects a sudden change in wind speed during the window adjustment period, causing a large area of ​​missed spraying and triggering an anomaly, the module applies an extension formula to the time weight of the current node. Forcibly extend the time resources of node two. The module will perform in-situ touch-up spraying within seconds. If a system logic fault (such as a sudden drop in spray pump pressure) or a sudden environmental change (such as a sudden strong wind of level 6 or above) is detected, the module will execute a topology reconstruction formula. Push subsequent nodes into the discard set, cut directed edges, skip the remaining operations, and directly realign the topology to the emergency return path to ensure the safety of the entire machine.

[0173] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for coordinated control of multi-task mounted equipment on unmanned aerial vehicles (UAVs), characterized in that, include: S1. Obtain the perception data of the work scenario, construct the work task chain with time weights, and extract the temporal features and control parameters of the current task node. S2. Control the multiple types of mounted equipment to execute the current task along the task chain, and trigger the window adjustment period when the remaining time weight of the current task node decreases to a preset threshold. S3. Collect operation effect data during the window adjustment period, and determine the current task chain flow status through effect verification logic; S4. Based on the aforementioned flow state, a feedforward blanking signal is sent to the flight control system to gradually blank the dynamic compensation amount corresponding to the current task, so as to smoothly transition the physical state. S5. Based on the physical characteristics of the load of the next task node to be executed, preload aerodynamic and center of mass change feedforward parameters into the flight control system, and generate a pre-activation command to be sent to the corresponding mounted equipment. S6. Monitor the actual feedback data during task execution and switching. When an abnormal state occurs, adaptively adjust the execution boundary of the task chain.

2. The method for coordinated control of multi-task mounted equipment for unmanned aerial vehicles according to claim 1, characterized in that, In S1, the task chain is implemented using a directed acyclic graph. The steps for constructing the task chain with time weights include: The task scenario is analyzed to deconstruct multiple task subtasks and the topological dependencies between them. The topological nodes and directed edges of the directed acyclic graph are generated based on the topological dependency mapping. Based on the time constraints of the job subtask, configure a first time weight representing the expected execution time for the topology node; Based on the UAV's maneuvering restrictions or payload switching delay, a second time weight representing the flow delay constraint is configured for the directed edge.

3. The method for coordinated control of multi-task mounted equipment for unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, In S1, the steps for extracting the temporal features and control parameters of the current task node include: The real-time execution progress of the job scenario is analyzed, and the remaining executable window of the current task node is calculated based on the time weight to generate the temporal features that characterize the node flow boundary. The operating status parameters and physical change data of the various types of mounted equipment are obtained, and the dynamic disturbance parameters of the whole system caused by load changes are simulated online through the airborne dynamics solution model.

4. The method for coordinated control of multi-task mounted equipment for unmanned aerial vehicles according to claim 1, characterized in that, In S2, the steps to trigger the window adjustment period include: The minimum convergence time is calculated as a preset threshold based on the load start-up and shutdown delay of the next task node to be executed and the current environmental interference. Monitor the dynamic completion rate of the current task and adjust the consumption rate of the remaining time weight of the current task node based on the deviation between the dynamic completion rate and the time already consumed. The remaining time weight is compared with the preset threshold in real time. When the remaining time weight decreases to the preset threshold, the window adjustment period is triggered.

5. The method for coordinated control of multi-task mounted equipment for unmanned aerial vehicles according to claim 1, characterized in that, In S3, the steps to determine the current task chain's transition state through effect verification logic include: Acquire real-time sensing data of the target area and extract residual attribute information of the target to be treated; After spatial registration, the feature data before and after the operation are compared, and a quantitative evaluation index characterizing the degree of elimination of the target to be treated is calculated. The quantitative evaluation index is compared with the preset safety threshold. If the threshold is met, a forward flow state variable is output to advance the task chain; otherwise, a reverse flow state variable is output to trigger node renewal.

6. The method for coordinated control of multi-task mounted equipment for unmanned aerial vehicles according to claim 1, characterized in that, In S4, the steps for gradually eliminating the dynamic compensation amount corresponding to the current task include: Based on the remaining duration of the window adjustment period, a time-varying blanking coefficient that monotonically decreases over time is constructed. The time-varying blanking coefficient is used to attenuate the dynamic compensation amount of the current task and generate a real-time attenuation control amount. The real-time attenuation control quantity is dynamically injected into the attitude control loop of the flight control system, and the dynamic compensation quantity is controlled to smoothly converge with the preload feedforward parameters of the next task to be executed at the end of the window adjustment period.

7. The method for coordinated control of multi-task mounted equipment for unmanned aerial vehicles (UAVs) according to claim 6, characterized in that, In S5, the step of preloading aerodynamic and center-of-gravity change feedforward parameters into the flight control system includes: Based on the physical characteristics of the load of the next task node to be executed, calculate the predicted value of the centroid offset and the predicted value of the aerodynamic parameter change. Based on the remaining duration of the window adjustment period, the predicted quantity is subjected to temporal smoothing mapping to generate a gradually increasing feedforward compensation sequence. During the window adjustment period, the feedforward compensation sequence is gradually injected, so that it is superimposed and fused with the real-time attenuation control quantity online.

8. The method for coordinated control of multi-task mounted equipment of unmanned aerial vehicles according to claim 1, characterized in that, In S6, the step of adaptively adjusting the execution boundary of the job task chain includes: When the abnormal state is that the task fails to meet the target, the time weight of the current task node is extended. When the abnormal state is a system logic failure or a sudden environmental change, the topology reconstruction of the job task chain is performed.

9. A collaborative operation platform for multi-task unmanned aerial vehicles (UAVs) with mounted equipment, characterized in that: The operating platform is used to execute the collaborative control method according to any one of claims 1-8, and comprises: The task chain configuration module is used to acquire the perception data of the work scenario, construct the task chain with time weights, and extract the temporal features and control parameters of the current task node. The timing coordination control module is used to control multiple types of mounted equipment to execute the current task along the task chain, and to trigger a window adjustment period when the remaining time weight of the current task node decreases to a preset threshold. The effect verification module is used to collect operation effect data during the window adjustment period and determine the current task chain flow status through effect verification logic. The feedforward blanking module is used to send a feedforward blanking signal to the flight control system according to the flow state, and to gradually blank the dynamic compensation amount corresponding to the current task so as to smoothly transition the physical state. The load preloading and activation module is used to preload aerodynamic and center of mass change feedforward parameters into the flight control system based on the load physical characteristics of the next task node to be executed, and generate a pre-activation command to be sent to the corresponding mounted equipment. The adaptive reconfiguration module is used to monitor the actual feedback data during task execution and switching. When an abnormal state occurs, it adaptively adjusts the execution boundary of the job task chain.

Images