A method for constructing a digital twin model of an unmanned aerial vehicle

Abstract

The present invention discloses a method for constructing a digital twin model of a drone, which belongs to the technical field of drone modeling. The design software is used to construct a digital twin three-dimensional geometric model of the drone, and a geometric twin mapping of the drone physical entity is obtained, which is saved and imported into the Unreal Engine to give the drone geometric model physical properties and behavioral functional characteristics; a drone physical frame structure model is constructed based on a simulation tool to reflect the drone behavioral characteristics, and a simulation actuator and a kinematic model are established based on the drone physical frame structure model based on the simulation tool, and the feedback data of the simulation actuator and the kinematic model are used to drive the motion of the drone geometric model in the Unreal Engine. The present invention realizes the mapping of physical entity objects in multiple dimensions such as geometric shape, physical properties and behavioral response, and has better twin performance.

Description

Technical Field

[0001] The present invention belongs to the technical field of unmanned aerial vehicle modeling, and particularly relates to a method for constructing a digital twin model of an unmanned aerial vehicle. Background Art

[0002] Unmanned aerial vehicles, commonly known as drones, are aircrafts controlled by ground equipment to complete flight missions and were initially applied in the military field. With the rapid development of micro-systems, artificial intelligence, and sensing technologies, quadrotor drones have also become popular. Due to their advantages such as small size, light weight, strong concealment, low cost, simple structure, and good flexibility, their application fields are constantly deepening and expanding. In addition to basic aerial photography, agricultural plant protection, daily inspections, police security, etc., quadrotor drones have also seen new developments in the fields of logistics distribution and emergency rescue. For example, drones equipped with life detectors can use the Doppler effect to sense vital signs for search and rescue; when there is a fire in the high altitude or high-rise buildings in the city, or when organizing rescue missions in mountains or at sea, drones can be used to drop life jackets, life ropes, or search for distress signals to trapped people for rescue; drones can be used to transport items such as virus samples, vaccines, and medicines, or carry disinfectants for disinfection. Thus, drones will be an essential high-tech backbone force in the future emergency rescue field.

[0003] Although the functions and application fields of drones are becoming more and more extensive, the environment faced during actual flight is becoming more complex. During the development of drones, it is difficult to conduct flight tests in real fire scenes, disaster areas, and complex environments with a large flow of people, and it takes a large amount of time. Therefore, there is an urgent need to develop a performance test simulation platform for drones in complex scenarios, which is of great significance for optimizing control algorithms and flight tests in complex application scenarios.

[0004] The UAV simulation platform is continuously iterated according to simulation requirements, and the functions achieved are also different. Currently, the proposed simulation platforms mainly include two categories: digital simulation platforms and hardware-in-the-loop simulation platforms. The digital simulation platform uses software to simulate various devices and environments, and the system operation process does not include any physical hardware. In 2018, considering the great significance of the UAV simulation platform for experimental teaching, Chen Jinyin et al. designed an open simulation platform for curriculum education based on the Robot Operating System (ROS). Although this platform is applicable to various algorithms, in order to improve the authenticity of data, it is necessary to collect environmental information in advance using acquisition devices such as Kinect to create the ROS environment required for the model, resulting in poor portability. In 2020, Tan Siyang et al. built a co-simulation environment based on the Matlab and Amesim platforms, and established a flight control law and an aircraft body model respectively for closed-loop simulation. Although communication between platforms can be achieved, there are still problems with poor real-time performance. The hardware-in-the-loop simulation platform adds some hardware to the simulation system, such as a flight control core board or relevant sensors, etc., and obtains real data through connection with the hardware and replaces some simulated data. The dSPACE real-time simulation system is developed by the German company dSPACE. By connecting with the RTW (Real Time Workspace) toolbox in MATLAB / Simulink, it has strong computing and code generation capabilities, and the speed of real-time simulation has been greatly improved. Major domestic universities and related research institutions have done a lot of research and development work on UAV real-time simulation systems based on real-time simulation devices such as dSPACE, but such platforms are very expensive and have a long development cycle. Bin Hu et al. designed a UAV hardware-in-the-loop simulation system based on rapid prototyping technology by combining xPC and Simulink technologies, shortening the R & D cycle. However, due to the lack of visual simulation, the intuitive visualization of data cannot be achieved. J. García et al. proposed a simulation platform and testing method based on the combination of a rotor UAV based on Pixhawk and Gazebo simulation. This platform demonstrates the control effect of simulated sensors and simulation models on UAV navigation and obstacle avoidance under real conditions. However, the modeling method uses the original mathematical modeling method, and the modeling means are single, resulting in the separation of the real-time operating conditions of physical entities and the data of the flight control model, and the availability of the collected information is low. Summary of the Invention

[0005] In order to solve the problems of the single traditional UAV modeling method, which is difficult to realize the mapping of the UAV digital model from multiple dimensions, resulting in the separation of the real-time data of physical entities from the model and low information availability, the present invention provides a method for constructing a UAV digital twin model, which realizes the mapping of physical entity objects in multiple dimensions such as geometric shape, physical properties, and behavioral responses, and has better twin performance.

[0006] The technical solution adopted by the present invention to solve its technical problems is as follows: A method for constructing a digital twin model of an unmanned aerial vehicle (UAV), including: constructing a three-dimensional geometric model of the digital twin of the UAV using design software to obtain the geometric twin mapping of the physical entity of the UAV, saving it and importing it into the Unreal Engine, and endowing the geometric model of the UAV with physical properties and behavioral functional characteristics; constructing a physical framework structure model of the UAV based on simulation tools to reflect the behavioral characteristics of the UAV, and based on the physical framework structure model of the UAV, establishing a simulation actuator and a kinematic model based on simulation tools, and driving the movement of the geometric model of the UAV in the Unreal Engine by the feedback data of the simulation actuator and the kinematic model.

[0007] As a further implementation of the present invention, the physical framework structure model of the UAV includes an actuator model, a kinematic / dynamic model, and a controller.

[0008] As a further implementation of the present invention, the actuator model is represented by a first-order inertial link:

[0009]

[0010] Where: τ is the time constant, Ω represents the angular velocity generated by the brushless DC motor, and Ω0 represents the initial angular velocity of the brushless DC motor.

[0011] As a further implementation of the present invention, the kinematic / dynamic model includes a UAV rotor dynamics model, a rigid body dynamics model, and a rigid body kinematic model.

[0012] The UAV rotor dynamics model is represented by the following formula:

[0013]

[0014] Where: T i represents the thrust generated by the i-th brushless DC motor, and Ω i represents the angular velocity generated by the i-th brushless DC motor; according to the rotation direction of the brushless DC motor, k represents the thrust factor k1 or the drag factor k2, which is determined by the following formula:

[0015]

[0016] Where, C T is the thrust coefficient, C P is the drag coefficient, ρ is the air density, and D is the propeller diameter.

[0017] As a further implementation of the present invention, the rigid body dynamics model is expressed as:

[0018]

[0019] Among them, U1 represents the total thrust generated by the propeller, U2 represents the total thrust when the UAV realizes roll motion, U3 represents the total thrust when the UAV realizes pitch motion, and U4 represents the total thrust when the UAV realizes yaw motion.

[0020] As a further embodiment of the present invention, the rigid body kinematic model is represented in the form of a combination of translational acceleration and rotational acceleration and is modeled as:

[0021]

[0022] Where: are the translational accelerations in three axial directions; φ, θ, ψ represent the roll angle, pitch angle, and yaw angle of the UAV in sequence; represent the roll angular velocity, pitch angular velocity, and yaw angular velocity of the UAV in sequence; represent the angular accelerations of the roll, pitch, and yaw motions of the UAV in sequence, m represents the mass of the UAV, g represents the acceleration due to gravity, l represents the length of the arm, and J r represents the moment of inertia of the brushless DC motor, and I x , I y and I z describe the moments of inertia about the x, y, and z axes respectively.

[0023] As a further embodiment of the present invention, the controller adopts a cascade control structure.

[0024] As a further embodiment of the present invention, the simulation tool communicates with the Unreal Engine through middleware to complete the digital mapping of the UAV behavior, so as to endow the UAV geometric model with dynamic behavior characteristics, that is, the UAV geometric model motion in the Unreal Engine is driven by the feedback data of the simulation actuator and the kinematic model; a fish-eye camera module is added to the middleware to sense the surrounding environment and feedback it to the digital twin model.

[0025] As a further embodiment of the present invention, the simulation actuator and the kinematic model include an actuator module, a kinematic and dynamic module, and a three-dimensional virtual simulation environment module; the input of the actuator module is the pulse width PWM command signal issued by the controller, as well as the UAV state information feedback in real time by the kinematic and dynamic module and the three-dimensional environment information sensed in the three-dimensional virtual simulation environment module, and the output is the pulling force and torque; the input of the kinematic and dynamic module is the reset signal and the pulling force and torque given by the actuator module, and the output is the UAV state information; the input of the three-dimensional virtual simulation environment module is the UAV state information, which visually displays the UAV state information, and the output is the three-dimensional environment information sensed by the fish-eye camera module.

[0026] As a further embodiment of the present invention, the UAV state information includes the UAV attitude and position information.

[0027] The beneficial effects of the present invention include: solving the problems of the single existing UAV modeling method and the separation of entity and model information, constructing a UAV digital twin model based on digital twin technology, realizing the mapping of the UAV physical entity object in multiple dimensions such as geometric shape, physical properties, and behavioral responses, and having better twin performance. Description of the Drawings

[0028] Figure 1 a is a geometric model diagram of the propeller of a quadcopter F450 UAV;

[0029] Figure 1 b is a geometric model diagram of the whole quadcopter F450 UAV;

[0030] Figure 2 is a model diagram of the physical frame structure of the UAV;

[0031] Figure 3 is a cascade control structure diagram of the controller;

[0032] Figure 4 is a simulation actuator and kinematic model diagram;

[0033] Figure 5 is a middleware model diagram;

[0034] Figure 6 is a structure diagram of the simulation experiment platform;

[0035] Figure 7 is a graph of the actual and desired attitude angles during flight;

[0036] Figure 8 is a diagram of the UAV motion state at the 40 - second moment;

[0037] Figure 9 is a system architecture diagram of the UAV flight simulation platform in Embodiment 1. Detailed Embodiments

[0038] Next, the technical solutions of the present invention will be clearly and completely described in conjunction with the accompanying drawings. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0039] In the description of the present invention, it should be noted that the orientation or positional relationship indicated by the terms "vertical", "horizontal", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or component referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation to the present invention. In addition, the terms "first", "second", "third" are only used to distinguish components and cannot be construed as indicating or implying relative importance.

[0040] In addition, the technical features involved in different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0041] Embodiment 1

[0042] A method for constructing a digital twin model of an unmanned aerial vehicle (UAV) studies the method for constructing a digital model of a UAV from three aspects: geometry, physics, and behavior in order to solve the problems of single UAV modeling method and separation of entity and model information. Based on digital twin technology, a digital twin model of the UAV is constructed to present the physical entity in a digital way, realizing the mapping of the physical entity object in multiple dimensions such as geometric shape, physical properties, and behavioral response.

[0043] 1. Geometric model

[0044] In this embodiment, a four-rotor F450 UAV is taken as the object, and its digital twin three-dimensional geometric model is constructed using CATIA software. Preferably, in order to solve the problem of modeling the propeller and the arm surface, surface modeling technology is used to establish a four-rotor propeller geometric model as shown in Figure 1 (a), which is consistent with the physical entity object in terms of appearance, size, and shape. Similarly, the model accessories such as the fuselage, arms, and landing gear are respectively completed, and they are assembled according to the actual size and geometric positional relationship to obtain the geometric twin mapping of the four-rotor F450 UAV physical entity as shown in Figure 1 (b). It is saved as an STL format file and imported into the virtual engine UE4 (Unreal Engine 4). By setting a new skeleton hierarchy through blueprint classes, the problem of attribute calibration is solved, and physical properties and behavioral functional characteristics are given to the F450 geometric model.

[0045] 2. Physical and mathematical model

[0046] To realize the behavior mapping of the F450 model, in this embodiment, a four-rotor UAV physical framework structure model that can reflect the behavioral characteristics of the UAV is designed, including an actuator model, a kinematic / dynamic model, and a controller model, as shown in Figure 2 shown.

[0047] The UAV actuator mainly plays the role of converting the PWM signal of the controller into the driving force of the motor, and its physical and mathematical model can be simplified as a first-order inertial link representation:

[0048]

[0049] Where: τ is the time constant, Ω represents the angular velocity generated by the brushless DC motor, and Ω0 represents the initial angular velocity of the brushless DC motor.

[0050] The kinematic / dynamic model includes the UAV rotor dynamics model, the rigid body dynamics model, and the rigid body kinematics model. The UAV rotor dynamics model is represented by Equation (2):

[0051]

[0052] Where: T i represents the thrust generated by the i-th brushless DC motor, and Ω i represents the angular velocity generated by the i-th brushless DC motor; according to the rotation direction of the brushless DC motor, k represents the thrust factor k1 or the drag factor k2, which is determined by Equation (3):

[0053]

[0054] Where, C T is the thrust coefficient, C P is the drag coefficient, ρ is the air density, and D is the propeller diameter. To objectively reflect the behavior and design performance of the propeller physical entity, this embodiment uses the F450 physical entity data as the modeling data, as shown in Table 1:

[0055] Table 1 Propeller characteristic table

[0056]

[0057] The thrust required for the UAV's ascent, roll, pitch, and yaw motions is determined by the speeds of the four brushless DC motors and can be modeled as:

[0058]

[0059] Where, U1 represents the total thrust generated by the propeller, U2 represents the total thrust when the UAV performs a roll motion, U3 represents the total thrust when the UAV performs a pitch motion, and U4 represents the total thrust when the UAV performs a yaw motion.

[0060] The four-rotor rigid body kinematics model is described in a combined form of translational acceleration and rotational acceleration and can be modeled as:

[0061]

[0062] Where: are the translational accelerations in three axial directions; φ, θ, ψ represent the roll angle, pitch angle, and yaw angle of the UAV in sequence; represent the roll angular velocity, pitch angular velocity, and yaw angular velocity of the UAV in sequence; represent the angular accelerations of the roll, pitch, and yaw motions of the UAV in sequence. m represents the mass of the UAV, g represents the acceleration due to gravity, l represents the length of the arm, and J r represents the moment of inertia of the brushless DC motor, and I x 、I y and I z describe the moments of inertia about the x, y, and z axes respectively.

[0063] To improve the system response speed and reduce the impact of the excessive lag of integration on the system response performance, the controller in this embodiment adopts the cascade control structure as shown in Figure 3 . By using the outer-loop proportional P control to accelerate the attitude adjustment and the inner-loop PID control to eliminate the oscillation and overshoot phenomena during the flight of the quadrotor UAV, the system robustness is improved.

[0064] 3. Digital Mapping of F450 UAV Behavior

[0065] According to the physical framework structure model of the quadrotor UAV, a simulation actuator and kinematic model as shown in Figure 4 is established by using the Aerospace Blockset and UAV Toolbox toolboxes in Simulink, including an actuator module, a kinematic and dynamic module, and a 3D virtual simulation environment module.

[0066] To endow the F450 UAV geometric model with dynamic behavior characteristics, this embodiment designs a middleware as shown in Figure 5 based on C++ dynamic DLL communication to solve the communication interface problem between Simulink and UE4, and realize the digital mapping of F450 UAV behavior, that is, the motion of the F450 UAV geometric model in UE4 is driven by the feedback data of the simulation actuator and kinematic model in Figure 4 .

[0067] To complete the immersive interaction between the digital model and the virtual environment in the UAV simulation flight experiment, a fisheye camera module is added to the middleware to sense the surrounding environment and feedback it to the twin model, realizing the interaction between the UAV and the 3D environment.

[0068] The input of the actuator module is the pulse-width PWM command signal sent by the controller, as well as the UAV state information feedback in real time by the kinematic and dynamic module and the 3D environment information sensed in the 3D virtual simulation environment module, and the output is the pulling force and torque;

[0069] The inputs of the kinematics and dynamics module are the reset signal, the pulling force and torque given by the actuator module, and the output is the state information of the UAV;

[0070] The input of the three-dimensional virtual simulation environment module is the state information of the UAV, which visually displays the state information of the UAV, and the output is the three-dimensional environment information sensed by the fisheye camera module;

[0071] In the above implementation, the UAV state information includes the UAV attitude and position information.

[0072] This embodiment is applied to the UAV flight simulation platform. From the initial physical entity data acquisition to the display and interaction of the service layer, the UAV flight simulation platform described in this embodiment consists of a physical layer, a virtual layer, a transmission layer, and a service layer, as Figure 9 shown.

[0073] The physical layer is the foundation of the entire platform and the only source for obtaining real sensing data. In this embodiment, Pixhawk is used to sense information such as attitude and position and transmit it to the virtual layer in real time. The virtual layer is the core part of the entire platform, consisting of the UAV digital twin model constructed through the above implementation and the three-dimensional virtual environment model, realizing the accurate modeling of physical entities and the reproduction of real scenarios. The transmission layer is the bridge and link between the physical layer and the virtual layer. Through the transmission network and the defined data interface, the Socket communication scheme is used to realize the two-way transmission of data. The service layer is the interface for human-computer interaction, using the ground station to realize the system control and operation status reproduction functions.

[0074] Embodiment 2

[0075] This embodiment gives the experimental and performance analysis scheme.

[0076] 1. Experimental environment and configuration

[0077] To verify the effectiveness of the simulation platform, interactive test experiments and flight attitude control experiments of the UAV digital twin model in the virtual scenario were carried out. As Figure 6 shown, the simulation experiment platform consists of a Pixhawk autopilot, a UAV digital twin model, a three-dimensional virtual environment digital mapping model (UE4 twin three-dimensional virtual scenario), and an F450 UAV remote control and receiver. The PC configuration for running the UAV digital twin model and the three-dimensional virtual scenario is: Intel Core i5 8300H processor, 16G memory, and NVIDIA GeForce GTX 1060 graphics card.

[0078] 2. Digital twin flight attitude performance test

[0079] To verify the attitude tracking performance of the UAV digital twin, the roll angle error, pitch angle error, and yaw angle error between the desired attitude and the actual attitude are analyzed respectively. The results are as Figure 7 shown. From 0 to 20 seconds, only the throttle was increased to raise the altitude, and no attitude transformation was performed. The pitch angle, roll angle, and yaw angle were all 0°. During the flight from 20 to 120 seconds, the attitude angles were randomly adjusted multiple times. The actual pitch angle, roll angle, and yaw angle changed following the desired attitude curve, could reach the desired attitude angle within 1 s, and the overshoot was less than 2%, with almost no steady-state error.

[0080] During the test, the motion state of the UAV at the 40-second moment in Figure 7 was selected as a reference, and the real-time motion state display effect at this moment was verified in the 3D virtual scene. At this time, the roll angle was 15°, the pitch angle was 8°, and the yaw angle was 20°, as shown in Figure 8 the figure.

[0081] The above results show that the digital twin model of the quadrotor UAV can better follow the attitude control instructions of the remote controller. The virtual simulation attitude changes in line with the expected attitude curve. The information of the physical entity is consistent with that of the twin model, and the information utilization rate is relatively high. The attitude transformation can be carried out in real time in the 3D environment, verifying the effectiveness of the UAV digital twin model.

[0082] Example 3

[0083] The method for constructing the 3D virtual environment model described in Example 1:

[0084] Establishing a virtual world that can map the physical world is the cornerstone of UAV visual simulation. Therefore, in this example, a twin virtual model of the real flight scene is constructed. The most critical element in the construction of the virtual scene is the building. Using 3ds Max to construct the simulation object can obtain a relatively realistic effect, but the modeling speed is too slow. Therefore, this example proposes a multi-element integrated 3D visualization virtual scene modeling method. According to the principle of similar normalization of building characteristic attributes, buildings are divided into landmark buildings (such as libraries and stadiums) and non-landmark buildings (such as dormitory buildings and teaching buildings). Use 3ds Max to perform refined modeling on landmark buildings, extract the vector grid of the building from remote sensing images, and then use City Engine software to perform large-scale modeling on non-landmark buildings. Integrate the models of 3ds Max and City Engine in UE4 for multi-element integration.

[0085] 1. Refined modeling

[0086] First, the important buildings in the scene were modeled in detail. Taking the N campus library as an example, its CAD data was imported into 3ds Max. After capture, extrusion, chamfering, and insertion, the rectangle tool was used to outline the edge of the wall. Secondly, according to the height of the building's top surface and its structure, different materials and texture elements were added to different areas of the model to increase the texture and realism of the building's main body.

[0087] In order to reduce the redundancy and complexity of the model and improve the running speed of the model, this embodiment proposes a method for optimizing the number of surfaces. In the modeling process, the invisible inner surfaces in the splicing area of the connected buildings are removed to reduce the generation of invalid surfaces and avoid the redundancy generated by a large number of duplicated structured models. For horizontal and vertical structures, the number of Boolean operations is minimized to reduce the complexity of the model. In order to minimize the number of models, the fine threshold in the model modifier is optimized to improve the running speed of the model while ensuring the authenticity of the building.

[0088] 2. Large-scale modeling

[0089] Large-scale modeling mainly solves the modeling of non-landmark buildings, trees and roads. To improve the modeling speed, this embodiment uses City Engine to perform large-scale vector data modeling based on the rule method. Traditional oblique photography technology uses the building roof as the standard for vector data extraction. The existing problems of building tilt, displacement and missing will lead to modeling deviations. This embodiment uses the bottom of the building as the standard and uses the FAME-Net network to train the aerial remote sensing image building data set to avoid the extraction deviation of the traditional method and extract the vector data of the building.

[0090] CGA (Computer Generated Architecture) rules are the core of building a large-scale modeling method, which focuses on modeling speed and efficiency, and can ignore some detailed information of buildings. To this end, modeling rules are written according to the structural type, floor height, and roof color of the building, and the corresponding type of buildings are generated in large quantities and quickly. The details of the building are related to the binding force of the rules. The more rules there are, the more complete the model details are.

[0091] In this embodiment, taking the N campus scenario as an example, in order to establish the CGA rule, it is necessary to find out the relationship between the number of floors and height of the building. The heights and numbers of floors of some buildings in the study area are measured, and Table 2 is drawn.

[0092] Table 2 Building height and number of floors

[0093]

[0094] According to Table 2, the relationship between building height and number of floors is obtained as shown in the following formula:

[0095] H = 3.46N + 0.69(6)

[0096] Specifically split the building into individual small structural components, conduct large-scale regular construction according to the building structure, floor height, and color, and use the texture mapping function to perform texture mapping on the doors, windows, roofs, and exterior walls of the building.

[0097] In addition, except for the building model, flowers, plants, trees, street lights, and road parts are established using existing rules to generate a three-dimensional virtual scene of N Campus.

[0098] 3. Model Multi-Fusion

[0099] To improve the immersion and interactivity of the simulation, import DEM (Digital Elevation Map) digital elevation data in UE4 for uneven terrain design. In this embodiment, the GDEMV2 elevation dataset is used as the original data source. However, a large amount of DEM data will affect the running speed of the subsequent virtual scene. At the same time, the data volumes describing flat and complex terrain areas are different, and the original DEM data needs to be processed. For this reason, this embodiment performs interpolation and noise reduction processing on the original data to improve data utilization. Then, import the terrain data into Global Mapper for three-dimensional expansion to obtain a terrain file in hfz format. After that, according to the width and height of the elevation map, set the resolution and data range in World Machine to obtain a RAW16 format height map file compatible with UE4. For multi-fusion of the scene in UE4, import the RAW16 format height map, select the material corresponding to the remote sensing image to create a real three-dimensional terrain with unevenness, import the buildings constructed by 3ds Max and CityEngine into UE4 in the same proportion, and place them on the constructed three-dimensional terrain. To solve the dynamic interaction between models, add collision settings between different objects to achieve the transformation of the scene from two-dimensional static to three-dimensional dynamic.

[0100] Example 4

[0101] Virtual Scene Immersion and Interaction Performance Test:

[0102] To conduct the immersion and interactivity tests of the virtual scene, let the drone digital model fly autonomously in the constructed main entrance scene of N Campus. Since the construction of the three-dimensional virtual scene is a twin mapping of the real scene, the mountains and terrain are consistent with the real environment, and elements such as trees and red flags in the scene will also move with the wind, and the whole scene has good immersion. When the drone senses an obstacle in the virtual scene, due to the collision settings in the scene, it will execute the action of avoiding the obstacle at that time, and has good interactivity with the environment, which can provide technical support for performance tests such as three-dimensional surveying and obstacle avoidance.

[0103] Obviously, the above embodiments are merely examples for clear illustration and are not limitations on the implementation manners. For those of ordinary skill in the art, other different forms of changes or variations can be made based on the above description. It is not necessary and impossible to list all implementation manners here. And the obvious changes or variations derived therefrom are still within the protection scope of the present invention.

Claims

1. A method for constructing a digital twin model of an unmanned aerial vehicle, characterized in that, Including: Construct a three-dimensional geometric model of the digital twin of the unmanned aerial vehicle (UAV) using design software to obtain the geometric twin mapping of the UAV physical entity, save it and import it into the Unreal Engine, and endow the UAV geometric model with physical properties and behavioral functional characteristics; construct a UAV physical framework structure model based on simulation tools to reflect the UAV behavioral characteristics, and based on the UAV physical framework structure model, establish a simulation actuator and kinematic model using simulation tools, and drive the movement of the UAV geometric model in the Unreal Engine by the feedback data of the simulation actuator and kinematic model; The UAV physical framework structure model includes an actuator model, a kinematic / dynamic model, and a controller; The actuator model is represented by a first-order inertial link: Where: τ is the time constant, Ω represents the angular velocity generated by the brushless DC motor, and Ω0 represents the initial angular velocity of the brushless DC motor; The kinematic / dynamic model includes a UAV rotor dynamics model, a rigid body dynamics model, and a rigid body kinematic model, The UAV rotor dynamics model is represented by the following formula: Where: T i represents the thrust generated by the i-th brushless DC motor, and Ω i represents the angular velocity generated by the i-th brushless DC motor; according to the rotation direction of the brushless DC motor, k represents the thrust factor k1 or the resistance factor k2, which is determined by the following formula: Among them, C T is the thrust coefficient, C P is the drag coefficient, ρ is the air density, and D is the propeller diameter; The rigid body dynamics model is expressed as: Where, U1 represents the total thrust generated by the propeller, U2 represents the total thrust when the UAV performs a roll motion, U3 represents the total thrust when the UAV performs a pitch motion, and U4 represents the total thrust when the UAV performs a yaw motion; The rigid body kinematic model is represented by a combination form of translational acceleration and rotational acceleration and is modeled as: Wherein: are the translational accelerations in three axial directions; φ, θ, ψ respectively represent the roll angle, pitch angle and yaw angle of the UAV; respectively represent the roll angular velocity, pitch angular velocity and yaw angular velocity of the UAV; respectively represent the angular accelerations of the roll, pitch and yaw motions of the UAV, m represents the mass of the UAV, g represents the gravitational acceleration, l represents the length of the arm, J r represents the moment of inertia of the brushless DC motor, I x , I y and I z respectively describe the moments of inertia about the x, y and z axes; The controller adopts a cascade control structure; The simulation tool communicates with the Unreal Engine through middleware to complete the digital mapping of the UAV behavior, so as to endow the UAV geometric model with dynamic behavioral characteristics, that is, the feedback data of the simulation actuator and kinematic model are used to drive the movement of the UAV geometric model in the Unreal Engine; a fish-eye camera module is added to the middleware to sense the surrounding environment and feedback it to the digital twin model; The simulation actuator and kinematic model includes an actuator module, a kinematic and dynamic module, and a three-dimensional virtual simulation environment module; the input of the actuator module is the pulse width PWM command signal sent by the controller, as well as the UAV state information feedback by the kinematic and dynamic module in real time and the three-dimensional environment information sensed in the three-dimensional virtual simulation environment module, and the output is the pulling force and torque; the input of the kinematic and dynamic module is the reset signal and the pulling force and torque given by the actuator module, and the output is the UAV state information; the input of the three-dimensional virtual simulation environment module is the UAV state information, visually displays the UAV state information, and the output is the three-dimensional environment information sensed through the fish-eye camera module; The UAV state information includes the UAV attitude and position information.

Images