Unmanned aerial vehicle flight control platform order sending method, equipment and medium

Abstract

The invention discloses an unmanned aerial vehicle flight control platform order dispatching method and device and a medium, and the method comprises the steps: constructing a hybrid model through training Wide and Deep; the Wide is used for processing continuous features and cross features of the unmanned aerial vehicle, and the Deep is used for processing sparse features of the unmanned aerial vehicle and the to-be-allocated task; when the task queue has the to-be-allocated task, inputting the state data of the plurality of unmanned aerial vehicles and the task information of the single to-be-allocated task into the hybrid model to obtain the adaptation degree between each unmanned aerial vehicle and the single to-be-allocated task; and sending the single to-be-allocated task to the target unmanned aerial vehicle with the highest adaptation degree. The unmanned aerial vehicle state data monitored in real time is fed back to the hybrid model composed of the Wide and the Deep, so that the task scheduling efficiency of the unmanned aerial vehicle and the intelligent allocation of task resources are improved.

Description

Technical Field

[0001] The present application relates to the technical field of unmanned aerial vehicle flight control and dispatching, and in particular to a method, device and medium for dispatching orders on an unmanned aerial vehicle flight control platform. Background Art

[0002] With the rapid development of science and technology, drone technology has become an important part of the modern scientific and technological field, and has shown great application potential in many fields such as logistics distribution, environmental monitoring, and post-disaster rescue.

[0003] However, with the increasing number of drones and the diversification of mission types, the existing flight control systems have gradually exposed many limitations in task allocation and flight scheduling.

[0004] At present, task allocation often relies on fixed rules or manual scheduling. When there are a large number of drones, the existing flight control system cannot make full use of drone status data and task information to make intelligent recommendations in task allocation. This leads to unreasonable task allocation, causing some drones to run out of power prematurely due to task overload, seriously affecting the smooth execution of the task and reducing the overall work efficiency and effectiveness. Summary of the invention

[0005] The embodiments of the present application provide a method, device and medium for dispatching orders for a UAV flight control platform, which are used to solve the problems of unreasonable task allocation and low efficiency in the dispatching process of the UAV flight control platform.

[0006] The present application embodiment adopts the following technical solutions: On the one hand, an embodiment of the present application provides a method for dispatching orders for a UAV flight control platform, the method comprising: constructing a hybrid model by training Wide and Deep; the Wide is used to process continuous features and cross features of the UAV, and the Deep is used to process sparse features of the UAV and the tasks to be assigned; when there are tasks to be assigned in the task queue, the status data of multiple UAVs and the task information of a single task to be assigned are input into the hybrid model to obtain the degree of fit between each UAV and the single task to be assigned; The single task to be assigned is dispatched to the target UAV with the highest adaptability.

[0007] In one example, the hybrid model is constructed by training Wide and Deep, specifically including: when jointly training the hybrid model, obtaining sample status data of the sample UAV and sample task information of the sample task to be assigned, as well as the sample adaptation probability between the sample status data and the sample task information; processing the continuous features and cross features in the sample status data through the Wide part to obtain the input features of the Wide part, and processing the sparse features in the sample status data and the sample task information through the Deep part to obtain the output features of the Deep part; inputting the input features of the Wide part and the output features of the Deep part into the joint training formula to obtain the adaptation probability between the sample UAV and the sample task to be assigned; calculating the difference between the sample adaptation probability and the adaptation probability through the loss function, so as to optimize the Wide weight, Deep weight and bias term of the joint training formula through the gradient descent method until the iteration is stopped.

[0008] In one example, when the task queue has tasks to be assigned, the status data of multiple drones and the task information of a single task to be assigned are input into the hybrid model to obtain the degree of fitness between each drone and the single task to be assigned, specifically including: when the task queue has tasks to be assigned, the flight status data of dispatched drones and / or the non-flight status data of undispatched drones and the task information of a single task to be assigned are input into the hybrid model to obtain the degree of fitness between each drone and the single task to be assigned.

[0009] In one example, after dispatching the single task to be assigned to the target UAV with the highest adaptability, the method further includes: determining the execution priority of the single task to be assigned; if the target UAV with the highest adaptability is an assigned UAV, searching the task list of the target UAV; the task list includes assigned tasks with different execution priorities; the higher the execution priority of the assigned task, the earlier it is executed; determining whether the execution priorities of the assigned tasks in the task list are different from the execution priorities of the single task to be assigned; if so, determining the execution order of the single task to be assigned according to the execution priority of the single task to be assigned, and inserting the single task to be assigned into the task list.

[0010] In one example, the method also includes: if there is a target assigned task with the same execution priority as the single task to be assigned, calculating a first distance difference between the task location of the target assigned task and the flight position of the target UAV; and calculating a second distance difference between the task location of the single task to be assigned and the flight position of the target UAV; when the first distance difference is less than the second distance difference, determining that the execution order of the target assigned task is earlier than the single task to be assigned; or, when the second distance difference is less than the first distance difference, determining that the execution order of the single task to be assigned is earlier than the single target assigned task.

[0011] In one example, after dispatching the single task to be assigned to the target UAV with the highest adaptability, the method further includes: if the target UAV is executing the single task to be assigned, based on the environmental information sent back by the target UAV, it is determined that the target UAV cannot continue to execute the single task to be assigned, and the single task to be assigned is recycled; when the target UAV has other assigned tasks, the target other assigned tasks whose execution order is second only to the single task to be assigned are determined, so that the target UAV executes the target other assigned tasks.

[0012] In one example, determining that the target drone cannot continue to execute the single task to be assigned based on the environmental information transmitted back by the target drone, and reclaiming the single task to be assigned, specifically includes: judging whether the transmitted environmental information includes that there is an obstacle blocking the execution of the single task to be assigned; if so, reclaiming the single task to be assigned; if not, determining that the target drone cannot reach the task location of the single task to be assigned based on the climate information in the transmitted environmental information and the flight status data of the target drone, and reclaiming the single task to be assigned.

[0013] In one example, after the single task to be assigned is recycled, the method further includes: marking the recycled single task to be assigned as abnormal, and aggregating the marked abnormal tasks into a table and sending it to the client.

[0014] On the other hand, an embodiment of the present application provides a drone flight control platform dispatching device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute any of the drone flight control platform dispatching methods described above.

[0015] On the other hand, an embodiment of the present application provides a non-volatile computer storage medium for dispatching unmanned aerial vehicle flight control platforms, which stores computer executable instructions, and the computer executable instructions can execute any of the above-mentioned unmanned aerial vehicle flight control platform dispatching methods.

[0016] At least one of the above technical solutions adopted in the embodiments of the present application can achieve the following beneficial effects: This application inputs the status data of drones and the task information of individual tasks to be assigned into a hybrid model of Wide and Deep. The recommendation algorithm intelligently generates the optimal task allocation plan based on multi-dimensional data input, avoiding the limitations of static rule systems. It realizes task allocation based on the real-time status of all drones, significantly improving the efficiency of task scheduling. This intelligent task allocation method can ensure that each drone is reasonably scheduled in a multi-drone environment, reduce the complexity and errors of manual drone scheduling, and ensure the smooth execution of each task. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solution of the present application, some embodiments of the present application will be described in detail below in conjunction with the accompanying drawings, in which: Figure 1 A schematic diagram of a flow chart of a method for dispatching orders on a UAV flight control platform provided in an embodiment of the present application; Figure 2 A schematic diagram of a hybrid model provided in an embodiment of the present application; Figure 3 A schematic diagram of the structure of a UAV flight control platform dispatching device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0019] Some embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0020] Figure 1 A flowchart of a method for dispatching orders on a UAV flight control platform provided in an embodiment of the present application. The method can be applied to different business fields. Certain input parameters or intermediate results in the process allow manual intervention and adjustment to help improve accuracy.

[0021] The analysis method involved in the embodiments of the present application can be implemented by a terminal device or a server, and the present application does not impose any special restrictions on this. For the convenience of understanding and description, the following embodiments are described in detail by taking the flight control platform server as an example.

[0022] Figure 1 The process in may include the following steps: S101: construct a hybrid model by training Wide and Deep; the Wide is used to process the continuous features and cross features of the drone, and the Deep is used to process the sparse features of the drone and the tasks to be assigned.

[0023] It should be noted that the Wide and Deep model is a model that combines the advantages of the traditional generalized linear model (Linear Model) and the deep neural network (DNN). It can process linear and nonlinear features at the same time, thus achieving a balance in memory and generalization capabilities, and is suitable for solving complex task scheduling problems.

[0024] In some embodiments of the present application, when jointly training the hybrid model, it is necessary to first obtain the sample status data of the sample drone and the sample task information of the sample task to be assigned, as well as the sample adaptation probability between the sample status data and the sample task information. The sample adaptation probability here is the true label value, usually 0 or 1, 0 represents that the sample task to be assigned is not executed by the sample drone, and 1 represents that the sample task to be assigned is executed by the sample drone.

[0025] The sample status data includes the flight status data and non-flight status data of the sample UAV, and the sample task information includes the task priority and sample task location of the sample task.

[0026] Then, the continuous features and cross features in the sample state data and the sample task information are processed by the Wide part to obtain the input features of the Wide part, and the sparse features in the sample state data and the sample task information are processed by the Deep part to obtain the output features of the Deep part.

[0027] It should be noted that continuous features include low-battery warning threshold, severe low-battery warning threshold, cumulative flight time of the drone, remaining flight time, and target longitude and latitude. These features reflect the basic status and historical performance of the drone. At the same time, the Wide part also captures the interaction between features through cross-feature generation. In addition, sparse features include task priority, drone operating status (flying status or non-flying status), current environmental information, etc. These features change with the operation of the drone and the execution of the task. Through the embedding layer, the sparse features are first mapped to a low-dimensional continuous vector space, and then complex feature learning is performed through a deep neural network. The Deep part improves the generalization ability of the model by capturing the nonlinear relationship between features, thereby making more accurate task predictions in different flight missions and environments.

[0028] The process of processing the continuous features and cross features in the sample state data and sample task information through the Wide part is as follows: The Wide part performs matrix mapping on the continuous features and cross features in the sample state data. The Wide part also captures the interaction between cross features through cross feature generation. For example, cross features such as "the intersection of drone battery power and remaining flight time" can help the model better understand the relationship between drone status and tasks.

[0029] It should be noted that the cross feature is generated by multiplying two features or other combinations.

[0030] The process of processing the sparse features in the sample state data and sample task information through the Deep part is as follows: The deep part first maps sparse features (i.e., discrete features) to a low-dimensional continuous vector space through an embedding layer. The embedding layer can convert high-dimensional sparse features into low-dimensional dense features, which makes these features more suitable for neural network training.

[0031] For example, the task priority can be converted into a low-dimensional continuous vector through the embedding layer, and the features processed by the embedding layer will be input into the deep neural network. The deep neural network can learn the complex relationship between features through multiple layers of nonlinear transformation. Specifically, the Deep part further learns and extracts the embedded features through multiple layers of fully connected layers.

[0032] Then, the input features of the Wide part and the output features of the Deep part are input into the joint training formula to obtain the predicted adaptation probability between the sample UAV and the sample to-be-assigned task. It should be noted that the adaptation probability is determined as the degree of adaptation.

[0033] The above joint training formula is as follows:

[0034] in, It is the sigmoid function, which is usually used for binary classification tasks and represents probability; It is the input feature of the wide part; It is the output of the Deep part and is the processing result of the deep neural network; is the Wide weight, is the Deep weight.

[0035] Finally, the difference between the sample adaptation probability and the predicted adaptation probability is calculated through the loss function, so as to optimize the Wide weight, Deep weight and bias term of the joint training formula by the gradient descent method until the iteration is stopped.

[0036] Among them, the loss function is as follows:

[0037] Among them, L is the difference value, y is the true label, is the predicted value (model predicted fit probability value), D is the training set.

[0038] It should also be noted that the process of optimizing the Wide weight, Deep weight, and bias term of the joint training formula through the gradient descent method is as follows: First, determine the initial Wide weight, initial Deep weight and bias term in the joint training formula, input the input features of the Wide part and the output features of the Deep part into the joint training formula to obtain the predicted adaptation probability, input the predicted adaptation probability and the known true label value into the loss function, and obtain the difference between the predicted adaptation probability and the true label. By using the gradient descent method, calculate the gradient of the loss function corresponding to the Wide weight, Deep weight and bias term, respectively, and update the Wide weight, Deep weight and bias term in the joint training formula in the opposite direction of the corresponding gradient, and use the updated Wide weight, Deep weight and bias term for a new round of sample training. The step size in the gradient formula is determined by the preset learning rate of the hybrid model. The gradient steps of calculating the Wide weight, Deep weight and bias term multiple times and the steps of updating the Wide weight, Deep weight and bias term in the opposite direction are iterated until the numerical difference between two adjacent results of the loss function is less than 1%. Stop the iteration and obtain the final optimized Wide weight, Deep weight and bias term.

[0039] S102: When there are tasks to be assigned in the task queue, the status data of multiple drones and the task information of a single task to be assigned are input into the hybrid model to obtain the degree of compatibility between each drone and the single task to be assigned.

[0040] In some embodiments of the present application, when there are tasks to be assigned in the task queue, the flight status data of the dispatched drones and / or the non-flight status data of the undispatched drones and the task information of the single task to be assigned are input into the hybrid model to obtain the degree of fit between each drone and the single task to be assigned. By inputting the flight status data of the dispatched drones and / or the non-flight status data of the undispatched drones and the task information of the single task to be assigned into the hybrid model, the optimal allocation route is obtained, which significantly improves the efficiency of task scheduling. This intelligent task allocation can ensure that each drone is reasonably scheduled in a multi-drone environment, reduce the complexity and errors of manual scheduling, and ensure the smooth execution of tasks.

[0041] It should be noted that the flight status data of the dispatched drones include battery power, remaining flight time, current flight speed, flight altitude, and location information. The drones exchange data with the central control platform through built-in sensors and transmit information that has been cleaned and normalized. Specifically, the location information is obtained through the GPS module. The battery management system (BMS) monitors the battery power in real time and uploads the drone's status information to the control platform through the wireless communication module.

[0042] It should be noted that at time point A, after one or more pending tasks in the task queue are dispatched, and within the preset time, there are no new tasks to be assigned, then all drones will execute their assigned tasks. Based on this, at time point B, when there are pending tasks in the task queue, the pending tasks at point B are new tasks relative to the pending tasks at point A.

[0043] Based on this, the execution priority of each task to be assigned is determined.

[0044] It should be noted that the execution priority is a numerical value starting from 0 and increasing by 1, with 0 being the highest priority; by determining the task priority, the execution order of each task is ensured, so that more important tasks can be assigned more quickly.

[0045] If the target drone with the highest adaptability is a dispatched drone, search the target drone's task list. The task list includes assigned tasks with different execution priorities. The higher the execution priority of the assigned task, the earlier it will be executed. For example, if there are two tasks with priority 4 and priority 5 in the task list, and the priority of the newly assigned task is 3, the newly assigned task with priority 3 will be executed first, and then the two tasks with priority 4 and priority 5 will be executed.

[0046] It should be noted that since the completed tasks are deleted from the task list, the assigned tasks in the task list are the assigned tasks that have not been executed by the target UAV.

[0047] Then, it is determined whether the execution priorities of the assigned tasks in the task list are different from the execution priorities of the single task to be assigned.

[0048] If so, the execution order of the single task to be assigned is determined according to the execution priority of the single task to be assigned, and the single task to be assigned is inserted into the task list.

[0049] If there is a target assigned task with the same execution priority as the single task to be assigned, calculate the first distance difference between the task location of the target assigned task and the flight position of the target UAV; and calculate the second distance difference between the task location of the single task to be assigned and the flight position of the target UAV.

[0050] When the first distance difference is less than the second distance difference, it is determined that the execution order of the target assigned task is earlier than the single task to be assigned; or, when the second distance difference is less than the first distance difference, it is determined that the execution order of the single task to be assigned is earlier than the single target assigned task.

[0051] It should be noted that, since there is a certain error in the distance value between each mission location and the drone, the mission locations of different missions will generally not have the same distance if the distance is accurately calculated.

[0052] It should be noted that if the target assigned task is the task that the target UAV is currently executing, the target assigned task will continue to be executed. After the target assigned task is completed, a single task to be assigned will be executed.

[0053] If the target drone with the highest adaptability is an unassigned drone, the unassigned drone directly executes a single task to be assigned.

[0054] S103: dispatching the single task to be assigned to the target UAV with the highest adaptability.

[0055] In some embodiments of the present application, after a single task to be assigned is dispatched to the target UAV with the highest adaptability, if the target UAV is executing the single task to be assigned, it is determined based on the environmental information sent back by the target UAV that the target UAV cannot continue to execute the single task to be assigned, and the single task to be assigned is recycled; when the target UAV has other assigned tasks, the execution order of other target assigned tasks that is second only to the single task to be assigned is determined, so that the target UAV executes the other target assigned tasks.

[0056] It should be noted that environmental information is obtained by collecting information on the climate conditions, wind speed, air pressure, temperature and humidity of the flight area in real time through environmental sensors (such as meteorological sensors and obstacle detection sensors), and by using lidar and millimeter-wave radar technology to detect obstacle information in the flight area.

[0057] It should be noted that environmental sensors (such as DHT22, BMP180) exchange data with the central platform through wired or wireless communication modules to update changes in the flight environment in real time.

[0058] After receiving the environmental information sent back by the target UAV, determine whether the sent back environmental information includes obstacles blocking the execution of the single task to be assigned. If so, recycle the single task to be assigned; if not, determine that the target UAV cannot reach the task location of the single task to be assigned based on the climate information in the sent back environmental information and the flight status data of the target UAV, and recycle the single task to be assigned. The recycle single task to be assigned is marked as abnormal, and the marked abnormal tasks are summarized into a table and sent to the client. By recycling abnormal tasks, users can better understand the impact of the current environment on the task and can help users better plan the subsequent task plan.

[0059] It should be noted that although the embodiments of the present application are based on Figure 1 Steps S101 to S103 are described in sequence, but this does not mean that steps S101 to S103 must be performed in a strict order. Figure 1 The order shown in the figure is to introduce and explain step S101 to step S103 in sequence, in order to facilitate those skilled in the art to understand the technical solution of the embodiment of the present application. In other words, in the embodiment of the present application, the order between step S101 to step S103 can be appropriately adjusted according to actual needs.

[0060] pass Figure 1 The method uses the state data of the UAV and the task information of a single task to be assigned to the hybrid model of Wide and Deep. The recommendation algorithm intelligently generates the optimal task allocation plan based on the multi-dimensional data input, avoiding the limitations of the static rule system. It realizes task allocation based on the real-time status of the UAV, significantly improving the efficiency of task scheduling. This intelligent task allocation can ensure that each UAV is reasonably scheduled in a multi-UAV environment, reducing the complexity and error of manual scheduling and ensuring the smooth execution of tasks.

[0061] Figure 2 A schematic diagram of a hybrid model provided in an embodiment of the present application.

[0062] exist Figure 2 The structure logic of the hybrid model is shown in the figure. The remaining flight time, longitude and latitude of the drone, and longitude and latitude of the target are directly input into the Concatenated Embeddings (1200 dimensions). The current state of the drone (flying state and non-flying state) and the task level (task priority) are input into the Concatenated Embeddings (1200 dimensions) through the embeddings layer. The connection vector is then input into three ReLU layers and the final Logistic Loss output unit.

[0063] Figure 3 A schematic diagram of a UAV flight control platform dispatching device provided in an embodiment of the present application includes: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute any of the above-mentioned methods for dispatching orders for a drone flight control platform.

[0064] Some embodiments of the present application provide a non-volatile computer storage medium for dispatching an unmanned aerial vehicle flight control platform, which stores computer-executable instructions, and the computer-executable instructions can execute any of the above-mentioned methods for dispatching an unmanned aerial vehicle flight control platform.

[0065] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device and medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0066] The devices and media provided in the embodiments of the present application correspond one-to-one to the methods. Therefore, the devices and media also have similar beneficial technical effects as the corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.

[0067] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0068] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0069] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0070] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0071] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0072] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM), and non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0073] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0074] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0075] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the technical principle of the present application should all fall within the protection scope of the present application.

Claims

1. A method for dispatching orders from a UAV flight control platform, characterized in that: The method comprises: By training Wide and Deep, a hybrid model is constructed; the Wide is used to process the continuous features and cross features of the drone, and the Deep is used to process the sparse features of the drone and the tasks to be assigned; When there are tasks to be assigned in the task queue, the status data of multiple drones and the task information of a single task to be assigned are input into the hybrid model to obtain the degree of fit between each drone and the single task to be assigned; The single task to be assigned is dispatched to the target UAV with the highest adaptability.

2. The method according to claim 1, characterized in that The hybrid model is constructed by training Wide and Deep, specifically including: When jointly training the hybrid model, sample state data of the sample UAV and sample task information of the sample task to be assigned, as well as a sample adaptation probability between the sample state data and the sample task information are obtained; Processing the continuous features and the cross features in the sample state data through the Wide part to obtain the input features of the Wide part, and processing the sparse features in the sample state data and the sample task information through the Deep part to obtain the output features of the Deep part; Inputting the input features of the Wide part and the output features of the Deep part into a joint training formula to obtain the adaptation probability between the sample UAV and the sample task to be assigned; The difference value between the sample adaptation probability and the adaptation probability is calculated through the loss function, so as to optimize the Wide weight, Deep weight and bias term of the joint training formula through the gradient descent method until the iteration is stopped.

3. The method according to claim 1, characterized in that When there is a task to be assigned in the task queue, the status data of multiple drones and the task information of a single task to be assigned are input into the hybrid model to obtain the degree of fit between each drone and the single task to be assigned, specifically including: When there are tasks to be assigned in the task queue, the flight status data of the assigned drones and / or the non-flight status data of the unassigned drones and the task information of the single task to be assigned are input into the hybrid model to obtain the degree of compatibility between each drone and the single task to be assigned.

4. The method according to claim 3, characterized in that After dispatching the single task to be assigned to the target UAV with the highest degree of adaptability, the method further includes: Determining the execution priority of the single task to be assigned; If the target drone with the highest adaptability is a dispatched drone, search the task list of the target drone; the task list includes assigned tasks with different execution priorities; the higher the execution priority of the assigned task, the earlier it will be executed; Determine whether the execution priority of the assigned tasks in the task list is different from the execution priority of the single task to be assigned; If so, the execution order of the single task to be assigned is determined according to the execution priority of the single task to be assigned, and the single task to be assigned is inserted into the task list.

5. The method according to claim 4, characterized in that The method further comprises: If there is a target assigned task with the same execution priority as the single task to be assigned, calculating a first distance difference between the task location of the target assigned task and the flight position of the target UAV; and calculating a second distance difference between the task location of the single task to be assigned and the flight position of the target UAV; When the first distance difference is less than the second distance difference, it is determined that the execution order of the target assigned task is earlier than the single task to be assigned; or, when the second distance difference is less than the first distance difference, it is determined that the execution order of the single task to be assigned is earlier than the single target assigned task.

6. The method according to claim 1, characterized in that After dispatching the single task to be assigned to the target UAV with the highest degree of adaptability, the method further includes: If the target drone is executing the single task to be assigned, it is determined based on the environmental information sent back by the target drone that the target drone cannot continue to execute the single task to be assigned, and the single task to be assigned is recycled; When the target UAV has other assigned tasks, determine the target other assigned tasks whose execution order is second only to the single task to be assigned, so that the target UAV executes the target other assigned tasks.

7. The method according to claim 6, characterized in that The step of determining, based on the environmental information transmitted back by the target drone, that the target drone cannot continue to execute the single task to be assigned, and recycling the single task to be assigned specifically includes: Determining whether the returned environmental information includes that there is an obstacle blocking the execution of the single task to be assigned; If so, reclaim the single task to be assigned; If not, based on the climate information in the returned environmental information and the flight status data of the target UAV, it is determined that the target UAV cannot reach the task location of the single task to be assigned, and the single task to be assigned is recovered.

8. The method according to claim 7, characterized in that After the single task to be assigned is recycled, the method further includes: The recovered single task to be assigned is marked as abnormal, and the marked abnormal tasks are summarized into a table and sent to the client.

9. A UAV flight control platform dispatching device, characterized in that: include: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the UAV flight control platform dispatching method described in any one of claims 1-8.

10. A non-volatile computer storage medium for dispatching orders for a UAV flight control platform, storing computer executable instructions, characterized in that: The computer executable instructions can execute a method for dispatching orders for a UAV flight control platform as described in any one of claims 1 to 8.

Images