An unmanned aerial vehicle (UAV) VR control training system and method simulating a dedicated controller drive

Abstract

The application relates to a simulation special controller driven unmanned aerial vehicle VR control practical training system and method, which comprises a trainee end, a special controller, a data conversion module and a virtual simulation platform. The special controller collects control input data and generates damping force according to force feedback control signals; the data conversion module generates flight control instructions, receives a flight data set and operation mode information to generate force feedback control signals; the virtual simulation platform obtains mode configuration information and generates operation mode information, receives flight control instructions, drives a virtual unmanned aerial vehicle to execute flight actions and obtains a flight data set. In summary, the application dynamically generates force feedback control signals according to the flight data set and the operation mode information through a closed-loop force feedback system, thereby realizing linkage between force feedback and flight states, and simultaneously supporting dynamic adjustment of force feedback characteristics according to the operation mode information, improving practical training pertinence and training effects.

Description

Technical Field

[0001] This application relates to the technical field of virtual simulation, and in particular to a VR control training system and method for unmanned aerial vehicles driven by a dedicated simulation controller. Background Technology

[0002] With the rapid development of the low-altitude economy and the continuous growth in demand for professional qualification certifications such as civil aviation drone pilot licenses and drone pilot 1+X vocational skills certificates, the drone operation training market is expanding, which puts forward higher requirements for the training efficiency, safety and standardization of assessment of training systems.

[0003] Traditional drone operation training uses real drones as the training medium, which has the problem of high training costs, requiring continuous investment in equipment purchase, venue rental and consumable consumption; at the same time, the flight safety risks are relatively high, and beginners are prone to equipment damage or even personal injury accidents due to operation errors; in addition, the progress of traditional training is constrained by external factors such as weather conditions and airspace control, making it difficult to achieve systematic training around the clock and at high frequency.

[0004] While existing drone control training systems based on virtual simulation technology can partially circumvent the aforementioned problems, they still have significant technical shortcomings: First, existing systems lack an effective force feedback closed-loop mechanism. The controllers they use typically do not have force feedback functionality, and the flight status information of the virtual drone cannot be converted into damping force output from the controller joystick. Trainees lack tactile feedback corresponding to real flight during operation, resulting in a disconnect between the simulated control experience and the real flight scenario, hindering the effective transformation of training results into real control capabilities. Second, the force feedback output in existing systems is decoupled from the operating mode and flight status. Even with a certain feedback mechanism, the feedback output is independent of the current operating mode, making it impossible to dynamically generate differentiated force feedback control signals based on real-time flight data and operating mode information. This results in a lack of targeted adjustments to the control damping characteristics experienced by trainees at different training stages, making it difficult to meet the needs of phased and differentiated training and assessment. Summary of the Invention

[0005] To address the aforementioned shortcomings, this application provides a VR control training system and method for unmanned aerial vehicles driven by a dedicated simulation controller.

[0006] The above-mentioned objective of this application is achieved through the following technical solution: A VR control training system for unmanned aerial vehicles driven by a simulation-specific controller includes a student terminal, a dedicated controller, a data conversion module, and a virtual simulation platform. The dedicated controller includes a force feedback unit and an interaction unit. The interaction unit is used to collect the control input data from the student's end and send it to the data conversion module. The force feedback unit is used to generate a corresponding damping force based on the force feedback control signal sent by the data conversion module. The data conversion module is used to generate flight control commands based on the control input data and send them to the virtual simulation platform; it is also used to receive the flight data set and operating mode information fed back by the virtual simulation platform, and generate force feedback control signals based on the flight data set and the current operating mode information. The virtual simulation platform is used to acquire mode configuration information and generate operating mode information based on the mode configuration information; it is also used to receive flight control commands, drive the virtual drone to perform corresponding flight actions, and acquire the corresponding flight data set. The flight dataset includes attitude angle data, velocity data, and motion angular velocity data; the operation mode information includes training mode information and assessment mode information; the data conversion module includes: The first conversion unit, when the operating mode information is training mode information, is used to calculate and generate eigencomponents based on attitude angle data, velocity data, and motion angular velocity data, and to calculate and generate guidance components based on the deviations between the attitude angle data and preset reference attitude angles and the deviations between the motion angular velocity data and preset reference angular velocities; wherein, the first conversion unit is further used to: When the virtual drone is determined to be in normal flight state, the intrinsic component and the guidance component are superimposed according to the preset first weight ratio to generate a force feedback control signal. When the virtual drone is determined to be in a dangerous flight state, the intrinsic component and the guidance component are superimposed according to the preset second weight ratio to generate a force feedback control signal. The second conversion unit, when the operating mode information is the assessment mode information, is used to generate a force feedback control signal that includes only intrinsic components based on the flight data set.

[0007] The second objective of this invention is achieved through the following technical solution: A VR control training method for drones driven by a dedicated simulation controller includes: The student inputs control input data through the interaction unit of the dedicated controller, and the interaction unit sends the control input data to the data conversion module; The data conversion module generates flight control commands based on the control input data and sends the flight control commands to the virtual simulation platform; The virtual simulation platform receives flight control commands, drives the virtual drone to perform corresponding flight actions, and acquires the corresponding flight data set; The virtual simulation platform feeds back flight data sets and operational mode information to the data conversion module; The flight data set includes attitude angle data, velocity data, and motion angular velocity data; the operation mode information includes training mode information and assessment mode information; the data conversion module generates a force feedback control signal based on the flight data set and the current operation mode information, and sends the force feedback control signal to a dedicated controller, specifically including: When the running mode information is training mode information, the eigencomponents are calculated based on the attitude angle data, velocity data and motion angular velocity data, and the guidance components are calculated based on the deviation between the attitude angle data and the preset reference attitude angle and the deviation between the motion angular velocity data and the preset reference angular velocity. When the virtual drone is determined to be in normal flight state, the intrinsic component and the guidance component are superimposed according to the preset first weight ratio to generate a force feedback control signal. When the virtual drone is determined to be in a dangerous flight state, the intrinsic component and the guidance component are superimposed according to the preset second weight ratio to generate a force feedback control signal. When the operating mode information is the assessment mode information, a force feedback control signal containing only intrinsic components is generated based on the flight data set. The force feedback unit of the dedicated controller generates a corresponding damping force based on the force feedback control signal, enabling the learner to perceive the flight status of the virtual drone through the damping force.

[0008] In summary, this application provides a VR control training system and method for unmanned aerial vehicles (UAVs) driven by a dedicated simulation controller. This system constructs a closed-loop force feedback system using a dedicated controller, a data conversion module, and a virtual simulation platform. It dynamically generates force feedback control signals based on flight data sets and operational mode information, enabling trainees to perceive the virtual UAV's state through touch and adapt to different training stages. This achieves real-time linkage between force feedback and flight status, allowing trainees to perceive the virtual UAV's flight status through touch. Furthermore, it supports dynamic adjustment of force feedback characteristics based on operational mode information, enhancing the training's relevance and effectiveness. Attached Figure Description

[0009] Figure 1 This is a schematic diagram of an embodiment of a VR control training system for unmanned aerial vehicles driven by a simulation-specific controller according to this application; Figure 2 This is an example diagram of an evaluation report in an embodiment of a VR control training system for unmanned aerial vehicles driven by a simulation-specific controller, as described in this application. Detailed Implementation

[0010] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0011] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0012] In one embodiment, such as Figure 1 As shown, this application discloses a VR operation training system for unmanned aerial vehicles driven by a simulation-dedicated controller, including a student terminal, a dedicated controller, a data conversion module, and a virtual simulation platform; The dedicated controller includes a force feedback unit and an interaction unit. The interaction unit is used to collect the control input data from the student's end and send it to the data conversion module. The force feedback unit is used to generate a corresponding damping force based on the force feedback control signal sent by the data conversion module. In this embodiment, the learner's end refers to the physical operation and perception end where the learner performs control training. It typically includes corresponding equipment and an interface connected to a dedicated controller, used to receive the learner's control commands and present feedback to the learner in a virtual environment. The dedicated controller is the device through which the learner interacts with the virtual drone. It integrates force feedback functionality, providing tactile feedback to the learner based on the virtual drone's flight status. The force feedback unit is a component of the dedicated controller. Its function is to generate corresponding damping force or tactile feedback on the controller based on the received force feedback control signal. The generated damping force is used to simulate the reaction force of a real drone on the joystick under different flight states, thereby enhancing the learner's immersion and realism. The interaction unit is another component of the dedicated controller. Its main function is to collect various control inputs from the learner on the controller, such as joystick displacement and button triggering. The control input data collected by the interaction unit is then sent to the data conversion module for processing. The control input data refers to the various operation commands issued by the learner to the virtual drone through the dedicated controller, such as throttle position, rudder deflection, pitch and roll angles, reflecting the learner's real-time control intentions. Force feedback control signal refers to the electrical or digital signal generated by the data conversion module based on the flight status and operation mode information of the virtual drone. It is sent to the force feedback unit of the dedicated controller to control it to generate a damping force of a specific magnitude and direction.

[0013] Specifically, the student-end is the interface through which the student operates and perceives the virtual environment. This can be a VR headset integrating a display screen and sensors to present an immersive virtual flight scene and track the student's head movements. The dedicated controller is the device for student input; it can be a physical device simulating the shape of a real drone remote controller, equipped with joysticks, buttons, and other control components. The student controls the virtual drone by operating these components. The dedicated controller further includes a force feedback unit and an interaction unit. The interaction unit collects the student's input data and sends it to the data conversion module. This interaction unit can consist of multiple physical joysticks and buttons, each integrating sensors such as potentiometers or Hall effect sensors to detect the student's displacement or pressing state and convert these analog signals into digital input data. The force feedback unit generates corresponding damping force based on the force feedback control signal sent by the data conversion module. This force feedback unit can consist of a miniature vibration motor installed inside the controller's joysticks; when it receives the force feedback control signal, the motor vibrates to simulate damping.

[0014] The data conversion module is used to generate flight control commands based on the control input data and send them to the virtual simulation platform; it is also used to receive the flight data set and operating mode information fed back by the virtual simulation platform, and generate force feedback control signals based on the flight data set and the current operating mode information. In this embodiment, the data conversion module processes the control input data from the trainee's end, converting it into flight control commands that the virtual simulation platform can recognize and execute. Simultaneously, the data conversion module also receives flight data and operating mode information fed back from the virtual simulation platform, and generates force feedback control signals accordingly to drive the force feedback unit of the dedicated controller. Flight control commands are commands generated by the data conversion module based on the trainee's control input data, instructing the virtual drone in the virtual simulation platform to perform specific flight actions, such as ascend, descend, forward, backward, left turn, right turn, etc. The flight data set refers to various flight state parameters of the virtual drone calculated and acquired by the virtual simulation platform in real time, such as attitude angle, speed, angular velocity, altitude, and position, reflecting the current flight status of the virtual drone. Current operating mode information is generated by the virtual simulation platform based on mode configuration information, indicating the current operating state of the training system, such as training mode, assessment mode, free flight mode, etc. Different operating modes may correspond to different force feedback generation logic and evaluation criteria. Mode configuration information refers to the parameters or instructions used to set the operating mode of the virtual simulation platform. These can be preset by the system administrator or trainees to determine which mode the system will use for simulation training or assessment.

[0015] Specifically, the data conversion module generates flight control commands based on the input data and sends them to the virtual simulation platform. This module can be a standalone hardware unit, such as an embedded processor, running a pre-defined algorithm that directly maps the user's input data (e.g., X / Y axis displacement of the joystick) to flight control commands for the virtual drone (e.g., pitch and roll angles). Alternatively, the module can be a software program running on a general-purpose computer, reading input interface data from a dedicated controller and converting it into flight control commands based on simple linear or nonlinear functional relationships. Furthermore, the module receives flight data sets and operating mode information from the virtual simulation platform and generates force feedback control signals based on these data sets and the current operating mode. The module can calculate and generate a fixed-strength force feedback control signal using a simple lookup table or a pre-defined scaling factor, based on the received flight data sets (e.g., the virtual drone's speed and attitude) and operating mode information (e.g., training mode). For example, as the virtual drone's speed increases, the force feedback signal strength also increases accordingly.

[0016] The virtual simulation platform is used to acquire mode configuration information and generate operating mode information based on the mode configuration information; it is also used to receive flight control commands, drive the virtual drone to perform corresponding flight actions, and acquire the corresponding flight data set. In this embodiment, the virtual simulation platform serves as the core computing and rendering unit. It is responsible for constructing the simulated virtual flight environment, mimicking the physical flight characteristics of the drone, receiving flight control commands, and driving the virtual drone to perform corresponding flight maneuvers. Furthermore, the virtual simulation platform generates and feeds back the virtual drone's flight data set and current operating mode information in real time. A virtual drone refers to a simulated drone entity within the virtual simulation platform. It possesses similar physical characteristics and flight dynamics models to real drones and can perform various flight maneuvers in the virtual environment according to flight control commands. Flight maneuvers refer to the various movements performed by the virtual drone in the virtual simulation platform according to flight control commands, such as takeoff, landing, hovering, level flight, turning, and rolling.

[0017] Specifically, the virtual simulation platform is used to acquire mode configuration information and generate operating mode information based on this information. The virtual simulation platform can be a high-performance computer system running specialized simulation software. Further, the virtual simulation platform can acquire mode configuration information, such as "training mode" or "free flight mode," by reading a configuration file, and then set the current operating mode accordingly. Alternatively, mode configuration information can be manually selected through the virtual simulation platform's user interface, set by the operator or trainee before each simulation start. In addition, the virtual simulation platform is used to receive flight control commands, drive the virtual drone to perform corresponding flight maneuvers, and acquire corresponding flight data sets. Internally, the virtual simulation platform can contain a basic physics engine. This engine calculates the virtual drone's trajectory and attitude changes in the virtual environment based on received flight control commands, such as "increase throttle," and generates flight data sets including position, velocity, and attitude in real time. For example, when receiving a "roll left" command, the physics engine simulates the drone tilting to the left and updates its attitude angle data.

[0018] For example, suppose user A wants to practice drone takeoff and hovering in a virtual environment. First, user A puts on a learning device, such as a VR headset, and enters the virtual flight scenario. At this time, the virtual simulation platform generates and sets the current operating mode information as the training mode according to preset mode configuration information, such as "beginner training mode".

[0019] User A inputs throttle control data via the interaction unit of a dedicated controller, for example, by pushing the left joystick upwards. The interaction unit collects the input data and sends it to the data conversion module. Upon receiving the throttle input data, the data conversion module converts it into flight control commands, such as "virtual drone ascends," according to a preset mapping relationship. The flight control commands are then sent to the virtual simulation platform. Upon receiving the "virtual drone ascends" flight control command, the virtual simulation platform drives the virtual drone to perform ascending flight maneuvers within the virtual environment. During the virtual drone's ascent, the virtual simulation platform calculates and acquires real-time flight data sets, such as current altitude, ascent speed, and attitude angles. The virtual simulation platform then feeds back the real-time acquired flight data sets and current operating mode information to the data conversion module.

[0020] After receiving the flight data set and operating mode information, the data conversion module generates a force feedback control signal according to the preset force feedback generation logic. For example, when the virtual drone ascends at a relatively high speed, the data conversion module may generate a strong force feedback control signal to simulate the reaction force of air resistance on the joystick and send the force feedback control signal to the dedicated controller.

[0021] After receiving the force feedback control signal, the force feedback unit of the dedicated controller generates a corresponding damping force on the left joystick. User A, sensing the damping force through the left joystick, can intuitively perceive the current ascent status and resistance of the virtual drone, thereby adjusting the control force for more precise takeoff and hovering operations. For example, when User A feels increased joystick resistance, they will realize that the virtual drone may be ascending rapidly or encountering significant wind resistance, and will fine-tune the throttle input to maintain stability.

[0022] Therefore, through the above process, the control input data from the student's end is converted into the flight actions of the virtual drone in real time. The flight state of the virtual drone is then converted into the damping force on the dedicated controller through force feedback control signals, forming a complete closed-loop feedback process. This allows user A to obtain tactile feedback similar to real flight when conducting drone control training in a virtual environment, thereby improving the realism and effectiveness of the training.

[0023] Compared to traditional drone operation training, this system, through a virtual simulation environment, can effectively reduce the high costs of equipment procurement, venue rental, and consumable consumption in real drone training, while eliminating flight safety risks. Trainees can practice repeatedly in a risk-free environment, significantly improving training efficiency. Furthermore, this system is not limited by weather conditions or airspace control, enabling all-weather, high-frequency systematic training, thus solving the problem of traditional training being constrained by external factors.

[0024] Compared to existing drone control training systems based on virtual simulation technology, this solution addresses the technical problem of lacking an effective force feedback closed-loop mechanism. Specifically, existing systems typically use controllers without force feedback, resulting in a lack of tactile feedback for trainees during operation, creating a disconnect between the simulated control experience and real flight scenarios. This solution, through a force feedback unit in a dedicated controller, generates corresponding damping forces based on the virtual drone's flight status information. This allows trainees to perceive the virtual drone's flight status through touch, such as air resistance during ascent or inertial forces during turns. This real-time, dynamic force feedback mechanism enhances the realism and immersion of the simulated control, effectively compensating for the shortcomings of existing systems in tactile feedback, thereby promoting the transformation of training results into real control capabilities.

[0025] Furthermore, this solution addresses the technical challenge of decoupling force feedback output from operating modes and flight states in existing systems. In existing systems with a feedback mechanism, the feedback output is often independent of the operating mode, making it impossible to dynamically generate differentiated force feedback control signals based on real-time flight data and current operating mode information. This system's data conversion module receives flight data and operating mode information from the virtual simulation platform and generates force feedback control signals accordingly. This means the strength and characteristics of the force feedback can be dynamically adjusted based on the current training stage and the specific flight state of the virtual drone. Through this differentiated force feedback generation mechanism, the control damping characteristics experienced by trainees at different training stages can be specifically adjusted, thereby meeting the needs of phased and differentiated practical training assessments and improving the precision and personalization of training.

[0026] In summary, this embodiment achieves highly realistic training for UAV operation by constructing a closed-loop feedback system that includes a dedicated controller, a data conversion module, and a virtual simulation platform. This not only solves the cost and safety issues of traditional training but also surpasses existing virtual simulation systems in terms of force feedback mechanism and dynamic adaptability, providing a more realistic and efficient solution for UAV operation training.

[0027] The flight dataset includes attitude angle data, velocity data, and angular velocity data; the operation mode information includes training mode information and assessment mode information; the data conversion module includes: The first conversion unit, when the operating mode information is training mode information, is used to calculate and generate eigencomponents based on attitude angle data, velocity data, and motion angular velocity data, and to calculate and generate guidance components based on the deviations between the attitude angle data and preset reference attitude angles and the deviations between the motion angular velocity data and preset reference angular velocities; wherein, the first conversion unit is further used to: When the virtual drone is determined to be in normal flight state, the intrinsic component and the guidance component are superimposed according to the preset first weight ratio to generate a force feedback control signal. When the virtual drone is determined to be in a dangerous flight state, the intrinsic component and the guidance component are superimposed according to the preset second weight ratio to generate a force feedback control signal. The second conversion unit, when the operating mode information is the assessment mode information, is used to generate a force feedback control signal that includes only intrinsic components based on the flight data set.

[0028] In this embodiment, attitude angle data describes the orientation of the virtual UAV in three-dimensional space, typically including pitch, roll, and yaw angles. This data can be obtained in real-time through the physics engine within the virtual simulation platform, or by simulating the output of real UAV sensors (such as an inertial measurement unit, IMU). Velocity data describes the movement speed of the virtual UAV in three-dimensional space, typically including velocity components along the X, Y, and Z axes, reflecting the UAV's translational motion. This is crucial for simulating air resistance, inertial forces, and providing velocity-related force feedback. This data can be obtained through real-time calculation by the physics engine within the virtual simulation platform, or by simulating the output of a GPS or visual odometry system. Angular velocity data describes the rotational speed of the virtual UAV about its own axes in three-dimensional space, typically including angular velocity components about the pitch, roll, and yaw axes. This reflects the rate of attitude change of the UAV, and is crucial for simulating rotational inertia, angular damping, and providing force feedback related to attitude changes. This angular velocity data can be obtained through real-time calculation by the physics engine within the virtual simulation platform, or by simulating the output of a gyroscope. The training mode information in the operation mode information indicates that the system is currently in the student's learning and practice phase. In training mode, the system usually provides more assistance and guidance to help students master basic operating skills. The assessment mode information indicates that the system is currently in the phase of evaluating the student's operating skills. In assessment mode, the system usually reduces or eliminates assistance and guidance to more realistically simulate the actual flight environment and objectively evaluate the student's independent operating ability.

[0029] In this embodiment, the first conversion unit is a logic or physical module used to generate a force feedback control signal containing intrinsic and guiding components based on the flight data set in training mode. Its function is to provide trainees with force feedback that has educational guidance significance, helping them understand the flight physics of the UAV and correct operational deviations. The first conversion unit can be an independent software module running on the processor of the data conversion module, or it can be a specific algorithm flow within the data conversion module. The intrinsic components are calculated based on attitude angle data, velocity data, and motion angular velocity data, aiming to simulate the inherent drag or reaction force generated by the UAV in its current flight state due to aerodynamics, inertia, and other physical characteristics. This reflects the real physical flight experience of the UAV; for example, when the UAV flies at high speed, the dedicated controller will feel greater air resistance; when the UAV's attitude changes rapidly, the dedicated controller will feel the reaction force caused by inertia. The intrinsic components can map attitude angles, velocities, and angular velocities to corresponding force feedback values ​​using preset aerodynamic and inertial models. For example, velocity-related drag can be calculated by multiplying the square of the velocity by the drag coefficient, and angular velocity-related drag can be calculated by multiplying the angular velocity by the moment of inertia. The guide components are generated based on the deviations between the attitude angle data and preset reference attitude angles, and between the motion angular velocity data and preset reference angular velocities. They aim to provide additional force feedback to help trainees adjust the UAV's attitude and angular velocity to a preset ideal state. For example, when the UAV's attitude deviates from the target, the dedicated controller will generate a force to pull the attitude back to the target direction. The guide components can be calculated using a PID controller or other control algorithms based on the errors between the current attitude angle and the reference attitude angle, and the errors between the current motion angular velocity and the reference angular velocity, to determine the required correction force. For example, when the pitch angle is too large, the guide component will generate a downward force, prompting the trainee to push the dedicated controller forward to reduce the pitch angle.

[0030] In this embodiment, the second conversion unit is also a logical or physical module used to generate a force feedback control signal containing only intrinsic components based on the flight data set in assessment mode. Its purpose is to provide an unassisted physical flight experience to objectively assess the trainee's independent operational capabilities. The second conversion unit can also be an independent software module running on the processor of the data conversion module, or as a specific algorithm flow within the data conversion module.

[0031] For example, when a trainee practices hovering a drone in training mode, the virtual simulation platform continuously sends the virtual drone's attitude angle data (e.g., current pitch, roll, and yaw angles), velocity data (e.g., horizontal and vertical velocities), and motion angular velocity data (e.g., pitch, roll, and yaw angular velocities) to the data conversion module. At this time, the first conversion unit in the data conversion module first calculates the intrinsic components. For example, based on the virtual drone's current small drift velocity, it simulates air resistance, causing the joystick of the dedicated controller to feel a certain resistance when the trainee attempts to counteract the drift; based on the small sway angular velocity of the drone's attitude, it simulates inertial drag, causing the joystick to feel a reaction force when the trainee attempts to stabilize the attitude. Secondly, the guidance component is calculated. If the virtual drone's pitch angle deviates from the preset reference pitch angle (e.g., 0 degrees for level flight), the guidance component will generate a slight force on the joystick of the dedicated controller's pitch axis, pushing it towards the center position, thus prompting the learner to adjust the pitch. Similarly, if the yaw rate is not zero (indicating the drone is rotating), the guidance force will act on the joystick of the yaw axis, encouraging the learner to stop rotating. Furthermore, these two components are superimposed. For example, if the drone is drifting slightly forward (velocity data indicates forward velocity) and simultaneously pitching slightly (attitude angle data indicates positive pitch), the learner will feel a combined force: a general drag due to forward motion (intrinsic component), and a specific downward thrust acting on the pitch joystick (guidance component) to correct the pitch.

[0032] For example, when a trainee enters the assessment mode to perform a precise hovering task, the second conversion unit in the data conversion module will only calculate the intrinsic components. At this time, when the trainee manipulates the dedicated controller, they will only feel the inherent drag and inertia of the virtual drone, without any active guiding force to correct its deviation. If the drone drifts or tilts, the trainee must rely entirely on visual cues observed in the VR environment and apply learned skills to apply the correct control inputs to bring the drone back to the desired state. In this case, force feedback will accurately reflect the drone's physical state, such as the increased drag during rapid maneuvers or the inertia felt when stopping rotating, without any system-provided assistance.

[0033] Through the above technical solution, this application can provide corresponding differentiated force feedback control signals according to different operating modes. Specifically, in training mode, by combining intrinsic and guided components, trainees can not only perceive the real physical characteristics of the virtual drone but also obtain real-time, guided force feedback, effectively helping them correct operational deviations and accelerating skill learning and mastery. In assessment mode, only intrinsic components are provided to ensure the purity and objectivity of the force feedback, thereby enabling a more accurate assessment of the trainee's independent control ability and avoiding the influence of auxiliary functions on the assessment results. Through this phased force feedback mechanism, the teaching effect and assessment accuracy of the drone VR control training system can be improved, enabling trainees to master drone control skills and obtain objective assessment results.

[0034] In one embodiment, the first conversion unit is specifically used for: Real-time acquisition of obstacle distance data of virtual drones, and retrieval of preset reference attitude angles and reference angular velocities; The wind resistance parameters experienced by the virtual drone are obtained, and the eigencomponents are generated based on the attitude angle data, velocity data, motion angular velocity data and wind resistance parameters through a preset aerodynamic mapping model. Based on the deviations between the attitude angle data and the reference attitude angle, and the deviations between the motion angular velocity data and the reference angular velocity, the guide component is calculated and generated through a preset PID correction mapping model. When the obstacle distance data is greater than the preset safe distance threshold and the attitude angle data does not exceed the preset first tilt angle threshold, it is determined to be a normal flight state. When the obstacle distance data is not greater than the preset safe distance threshold, or the attitude angle data exceeds the preset first tilt angle threshold, it is determined to be a dangerous flight state.

[0035] In this embodiment, attitude angle data typically refers to Euler angles (such as roll, pitch, and yaw angles) or quaternions of the virtual UAV in space, used to describe its spatial attitude. Velocity data refers to the linear velocity components of the virtual UAV in three-dimensional space, reflecting its speed and direction of movement. Angular velocity data refers to the rotational speed of the virtual UAV around its own axis, reflecting the rate of attitude change. Obstacle distance data refers to the distance information between the virtual UAV and the nearest obstacle, which can be obtained through collision detection or ray detection mechanisms in the virtual simulation platform. Reference attitude angles and reference angular velocities are pre-set ideal or target attitudes and angular velocities, used to provide reference and correction during training; they can be stored in a configuration file or dynamically set by the training task. Wind drag parameters refer to the air resistance, wind force, and other environmental factors experienced by the virtual UAV during flight, which can be calculated or simulated in real time based on the wind field model in the virtual environment. An aerodynamic mapping model is a mathematical model used to simulate the aerodynamic forces and moments experienced by the UAV under different attitude, speed, and wind field conditions; for example, it can be a model based on lookup tables, polynomial fitting, or simplified physical equations. Deviation refers to the difference between the current attitude angle data and the reference attitude angle, and between the current motion angular velocity data and the reference angular velocity. It is usually calculated through subtraction or a complex error function. The PID correction mapping model is a feedback controller applied to control systems. It generates correction output by calculating proportional, integral, and derivative terms. Its function is to adjust the output according to the deviation value, making the system tend towards the target value. The guiding component refers to the auxiliary force in the force feedback control signal used to guide the trainee to correct their operation, making their flight attitude and trajectory approach the preset target. The safe distance threshold is a preset minimum safe distance. When the distance between the virtual drone and an obstacle is less than the safe distance threshold, a collision risk is considered. The first tilt angle threshold is a preset maximum safe attitude tilt angle. When the attitude angle of the virtual drone exceeds the first tilt angle threshold, it is considered to be in an unstable and dangerous state. Normal flight state refers to the flight state of the virtual drone within the safe distance and with a stable attitude. Dangerous flight state refers to the flight state of the virtual drone when it approaches an obstacle or exhibits abnormal attitude. The first and second weighting ratios are sets of coefficients used to adjust the proportions of the intrinsic and guided components in the force feedback control signal. These can be fixed values ​​or dynamically adjusted based on training difficulty or trainee performance. Superposition refers to combining the intrinsic and guided components, for example, through vector addition or weighted averaging, to generate the final force feedback control signal.

[0036] Furthermore, the first weighting ratio can be 70%-80% for intrinsic components and 20%-30% for guiding components, and the second weighting ratio can be 20%-30% for intrinsic components and 70%-80% for guiding components.

[0037] As an example, the first weighting ratio is 75% intrinsic component and 25% guidance component. This weighting ratio is suitable for normal flight conditions, ensuring that learners receive appropriate control guidance while gaining a realistic flight feel. The second weighting ratio is 25% intrinsic component and 75% guidance component. This weighting ratio is suitable for dangerous flight conditions, significantly enhancing the guidance component and prompting learners to make control corrections through obvious changes in joystick damping force, thus preventing the virtual drone from crashing or colliding.

[0038] Furthermore, as an example, the safe distance threshold can be set to 8 meters. When the distance between the virtual drone and an obstacle is less than 8 meters, it is determined to be a dangerous flight state, automatically switching to the second weight ratio to enhance the guidance component and prompting the user to avoid the obstacle. The emergency distance threshold can be set to 2 meters. When the distance between the virtual drone and an obstacle is less than 2 meters, emergency collision avoidance logic is triggered, locking the control input data facing the obstacle and forcibly performing a deceleration and hovering operation to avoid a collision. The static obstacle detection distance can be set to 10 meters. The detection distance for static obstacles (such as buildings, trees, and the ground) is set to 10 meters. When the distance between the virtual drone and a static obstacle is less than 10 meters, a collision warning is initiated. The dynamic obstacle detection distance can be set to 15 meters. The detection distance for dynamic obstacles (such as moving vehicles, pedestrians, and other drones) is set to 15 meters. Since the position of dynamic obstacles is constantly changing, a longer warning distance is needed to allow sufficient reaction time.

[0039] Furthermore, the calculation of the intrinsic components is based on a pre-defined aerodynamic mapping model, which simulates the aerodynamic feedback forces experienced by a real UAV during flight. The specific calculation process is as follows: ,in, The inertial force component is calculated as follows: , This is the inertial force coefficient, typically ranging from 0.1 to 0.3. The pitch angular velocity, For the roll angular velocity, The yaw rate is used to simulate the inertial feedback generated by the UAV during attitude changes. When the UAV performs rapid maneuvers (i.e., with a large angular velocity), the inertial force component increases, the joystick damping force increases, and the trainee can feel a significant heaviness. The air resistance component is calculated as follows: , This is the drag coefficient, typically ranging from 0.05 to 0.15. The velocity along the X-axis. The velocity along the Y-axis. Z-axis velocity, The pitch angle is used to simulate the air resistance encountered by the drone during flight. When the drone flies at high speed (high speed) or dives at a large angle (large pitch angle), the air resistance component increases, and the joystick damping force increases. The wind field drag component is calculated as follows: , This is the wind field influence coefficient, typically ranging from 0.02 to 0.08. For wind speed, Yaw angle The wind direction is represented by the wind drag component, which simulates the wind force affecting the UAV in a windy environment. When the UAV flies against the wind (heading is opposite to the wind direction), the wind drag component increases, and the stick damping force increases; when the UAV flies with the wind (heading is the same as the wind direction), the wind drag component decreases, and the stick damping force decreases.

[0040] For example, the first conversion unit can be deployed on a high-performance microcontroller within the data conversion module, such as an ARM Cortex-M series processor. This microcontroller receives flight data from the virtual drone in real-time from the virtual simulation platform via a high-speed data bus, updating attitude angle data, velocity data, motion angular velocity data, and obstacle distance data at a frequency of 100 frames per second. Simultaneously, the microcontroller retrieves the reference attitude angle and reference angular velocity required for the current training task from its internal memory. When calculating the intrinsic components, the microcontroller calls a pre-programmed aerodynamic mapping model, which may be a function based on lookup tables and linear interpolation. Based on the input attitude angle data, velocity data, motion angular velocity data, and wind resistance parameters provided by the virtual simulation platform, it quickly calculates the force feedback component simulating real flight drag. When calculating the guidance component, the microcontroller executes a preset digital PID controller algorithm, using the deviations between the current attitude angle data and the reference attitude angle, and the deviations between the motion angular velocity data and the reference angular velocity as inputs, to calculate the force feedback component used to correct the trainee's actions. Furthermore, the microcontroller determines the current flight state based on the received obstacle distance data (e.g., a safe distance threshold of 5 meters) and attitude angle data (e.g., a first tilt angle threshold of 45 degrees). If the obstacle distance is greater than 5 meters and the attitude angle does not exceed 45 degrees, it is considered a normal flight state. In this case, the microcontroller vector-superimposes the two components according to a first weighting ratio (e.g., intrinsic component weight 0.7, guide component weight 0.3). Conversely, if the obstacle distance is not greater than 5 meters or the attitude angle exceeds 45 degrees, it is considered a dangerous flight state. In this case, the microcontroller switches to a second weighting ratio (e.g., intrinsic component weight 0.4, guide component weight 0.6) for superposition, generating the final force feedback control signal and sending it to the dedicated controller.

[0041] Through the above technical solution, this application can dynamically adjust the proportion of intrinsic and guiding components in the force feedback control signal according to the real-time flight status of the virtual drone, enabling trainees to obtain more realistic and instructive tactile feedback during training. Especially when approaching dangerous situations, it can provide a stronger sense of corrective force, effectively guiding trainees to avoid potential collisions or loss of control risks. This not only enhances the immersion and realism of training, but also helps trainees quickly establish the ability to perceive dangerous situations and develop coping strategies in a safe and controllable virtual environment, thereby improving the efficiency and safety of mastering drone operation skills.

[0042] In one embodiment, the second conversion unit is specifically used for: Real-time acquisition of wind resistance parameters experienced by the virtual drone; Based on attitude angle data, velocity data, motion angular velocity data, and wind field drag parameters, eigencomponents are generated through a pre-set aerodynamic mapping model. The intrinsic components are directly output as force feedback control signals, and flight control commands are generated based on the control input data from the student's end.

[0043] In this embodiment, attitude angle data typically includes roll angle, pitch angle, and yaw angle, used to describe the attitude of the virtual UAV in three-dimensional space; velocity data represents the speed and direction of movement of the virtual UAV in space, such as the linear velocity components on the X, Y, and Z axes; motion angular velocity data reflects the rotational speed of the virtual UAV around its own axis, such as roll angular velocity, pitch angular velocity, and yaw angular velocity. These data can be calculated by the physics engine within the virtual simulation platform and sent to the data conversion module at high frequency. Wind drag parameters may include air density, the windward area of ​​the virtual UAV, and the drag coefficient, which together determine the magnitude and direction of air drag experienced by the virtual UAV during flight. The wind drag parameters can be dynamically calculated and provided by the virtual simulation platform based on the wind field model, airflow turbulence, and the real-time flight status of the virtual UAV in the simulated environment. The preset aerodynamic mapping model is a mathematical or physical model used to transform the flight state data and environmental parameters of a virtual drone into aerodynamic forces and moments acting on the drone. It can be constructed based on simplified aerodynamic equations, lookup tables, or complex computational fluid dynamics simulation results. Its purpose is to simulate the aerodynamic forces experienced by a real drone under different flight conditions, thus providing a physical basis for force feedback. Directly outputting the intrinsic components as force feedback control signals means that, in assessment mode, the force feedback control signal received by the dedicated controller force feedback unit consists only of intrinsic components, without any guiding or corrective components. This ensures that the damping force felt by the trainee during the assessment originates entirely from the real physical feedback of the virtual drone, thereby objectively evaluating the trainee's control ability in a purely physical environment. Generating flight control commands based on the trainee's control input data means that the second conversion unit receives the control input data input by the trainee through the dedicated controller interaction unit and converts it into flight control commands that the virtual simulation platform can recognize and execute. These commands directly drive the virtual drone to perform corresponding flight actions, such as gaining altitude, turning, diving, or rolling.

[0044] For example, the second conversion unit can be configured as an embedded processor or a high-performance microcontroller, such as an STM32 series chip based on the ARM architecture. It interacts with the virtual simulation platform via a high-speed serial communication interface (such as SPI or CAN bus) to receive real-time attitude angle data, velocity data, motion angular velocity data, and wind drag parameters from the virtual drone. This data is transmitted in floating-point or fixed-point format and timestamped to ensure consistency. A preset aerodynamic mapping model can be stored as a lookup table or a set of pre-compiled mathematical functions, such as an attitude dynamics model based on quaternions and a drag calculation model based on velocity vectors. Upon receiving flight data and wind drag parameters, the second conversion unit immediately performs model calculations to generate eigencomponents representing the drone's inertia, air drag, and other physical effects. For example, based on the virtual drone's current speed and frontal area, combined with air density and drag coefficient, the corresponding air drag torque can be calculated and used as part of the eigencomponents. Furthermore, the calculated intrinsic components are converted into analog voltage or current signals via a digital-to-analog converter, directly driving the force feedback motor in the dedicated controller's force feedback unit, thereby generating corresponding damping forces on the left and right joysticks. Simultaneously, analog signals input by the trainee via the left and right joysticks (e.g., voltage values ​​output by potentiometers) are acquired by an analog-to-digital converter and directly mapped into flight control commands for the virtual drone, such as throttle, yaw, pitch, and roll, and sent to the virtual simulation platform via a communication interface.

[0045] Through the above technical solutions, in the assessment mode, trainees can experience physical feedback close to that of real drone flight, without any assistance or guidance. This allows for accurate assessment of trainees' operational capabilities in a purely physical environment, ensuring the objectivity and fairness of the assessment results and providing reliable technical support for the skill assessment of drone pilots.

[0046] In one embodiment, the data conversion module further includes a control priority unit, which is used to: A control priority sequence is set, which includes emergency collision avoidance logic, fuselage extreme attitude protection, flight physical parameter limiting, auxiliary stabilization algorithm, and control input data processing in sequence. When the distance data of the detected obstacle is less than the preset emergency distance threshold, the emergency collision avoidance logic is triggered, locking the control input data in the direction of the obstacle and forcibly performing deceleration and hovering operation; When the attitude angle data exceeds the preset second tilt angle threshold, or the motion angular velocity data exceeds the preset angular velocity threshold, the fuselage extreme attitude protection is triggered, the control input data of the student terminal is cut off and the attitude return operation is forcibly executed. When several high-priority control logics are triggered simultaneously, they are executed sequentially according to the control priority sequence. The output of the high-priority control logic will override the output of the low-priority control logic. When the virtual drone is in normal flight, the intrinsic components and the guidance components are mixed according to the first weight ratio, and flight control commands are generated by combining the control input data.

[0047] In this embodiment, the control priority unit is a component used to manage and coordinate different control strategies or logics. The control priority unit can be a software module running on the processor of the data conversion module, responsible for receiving various sensor data and system status, and determining the control action to be executed based on preset rules and priority sequences. The control priority unit can be implemented using a state machine, rule engine, or priority-based scheduling algorithm. The control priority sequence is a sequence obtained by sorting various predefined control logics according to their importance and urgency. The control priority sequence can be stored in the memory of the control priority unit and dynamically queried and executed during system runtime. Its implementation can be an ordered list, array, or linked list structure, where each element represents a control logic and its corresponding priority.

[0048] In this embodiment, the emergency collision avoidance logic is a high-priority safety mechanism designed to prevent the virtual drone from colliding with obstacles. When a collision risk is detected, the emergency collision avoidance logic immediately intervenes to override the user's control input and enforces an evasive maneuver. This logic can be implemented through a separate algorithm module that continuously monitors obstacle distance data and outputs mandatory flight control commands, such as reverse thrust or lateral translation commands, when specific conditions are met. The fuselage extreme attitude protection is a mechanism to prevent the virtual drone from entering dangerous attitudes. When the virtual drone's attitude angle or angular velocity exceeds a safe range, the fuselage extreme attitude protection intervenes to correct the drone's attitude. This protection can be implemented through an attitude monitoring and correction algorithm that compares the current attitude data with a preset threshold in real time and generates a control command to correct the attitude when the threshold is exceeded. Flight physics parameter limiting is a mechanism to restrict the virtual drone's flight parameters to ensure they are physically reasonable and safe, such as limiting maximum speed, maximum acceleration, and maximum tilt angle. The flight physical parameter limiting can be achieved by checking and trimming the range of various parameters before generating flight control commands. The auxiliary stabilization algorithm is an algorithm that helps trainees control the drone more stably, such as attitude maintenance, altitude hold, and point hold. This algorithm can generate additional stabilizing control quantities based on the trainee's input and the drone's current state using a PID controller or other control algorithms. Control input data processing involves parsing, filtering, and mapping the raw control data input by the trainee through the interaction unit, converting it into flight control commands executable by the virtual drone. This processing can be achieved through a series of signal processing and mapping functions. Locking control input data towards obstacles and forcing deceleration and hovering is a specific emergency collision avoidance strategy. When the emergency collision avoidance logic is triggered, the system identifies the direction of the obstacle and prevents the trainee from controlling in that direction, while automatically issuing deceleration and hovering commands. This can be achieved by modifying the effective range of the control input data or directly overriding the trainee's input. Interrupting the user's input and forcibly performing attitude correction is a specific strategy for protecting the aircraft's extreme attitude. When this protection is triggered, the system ignores the user's input and automatically restores the UAV's attitude to a safe range. This can be achieved by setting the user's input to zero or directly generating an attitude correction command. The output of high-priority control logic overrides the output of low-priority control logic, a priority arbitration mechanism. When multiple control logics are triggered simultaneously, the system selects the output of the highest-priority logic as the final control command based on a preset priority sequence, ignoring or suppressing the outputs of other low-priority logics. This can be achieved using a multiplexer or a weighted summation method, where high-priority logics have a significantly greater weight than low-priority logics.The system generates flight control commands by mixing intrinsic and guiding components according to the first weight ratio. This means that under normal flight conditions, the system will generate the final flight control commands based on the student's control input and the mixing ratio of intrinsic and guiding components in the force feedback signal. This indicates that the system will still provide assistance and guidance in non-emergency situations.

[0049] For example, in a VR drone control training session, a trainee is controlling a virtual drone. During flight, if the obstacle distance data suddenly falls below a preset emergency distance threshold (e.g., the virtual drone is rapidly approaching a virtual building), the control priority unit detects this emergency and triggers emergency collision avoidance logic. This immediately locks control input data towards the obstacle; for example, if the obstacle is ahead, the trainee's instruction to push the joystick forward will be ignored or reversed. Simultaneously, a deceleration and hovering operation is forcibly executed, causing the virtual drone to automatically slow down and hover at a safe distance, avoiding a collision with the virtual building. In another scenario, while performing advanced maneuvers, the trainee might inadvertently cause the virtual drone's attitude angle data to exceed a preset second tilt angle threshold, or its motion angular velocity data to exceed an angular velocity threshold, bringing the virtual drone to the brink of loss of control. In this case, the control priority unit triggers the fuselage extreme attitude protection logic, immediately cutting off control input data from the trainee and forcibly executing an attitude recovery operation (e.g., automatically adjusting the rotor speed), causing the virtual drone to quickly return to a stable flight attitude, preventing it from rolling or crashing. In these emergency intervention scenarios, if multiple control logics are triggered simultaneously—for example, both collision avoidance and attitude protection requirements—the control priority unit will execute them sequentially according to their priority sequence. This ensures that the output of higher-priority control logic overrides the output of lower-priority control logic, thereby guaranteeing decision consistency and safety in complex situations. In normal flight conditions, the control priority unit generates flight control commands based on the input data, mixing intrinsic and guided components according to a first-weighted ratio. This provides trainees with a smooth flight experience accompanied by assisted guidance.

[0050] Through the aforementioned technical solution, this application introduces a hierarchical control priority mechanism. Specifically, when the virtual drone faces an emergency collision with an obstacle or its attitude reaches its limit, the control priority unit can promptly trigger high-priority control logic, such as emergency collision avoidance logic or extreme attitude protection, actively locking control input data towards the obstacle or cutting off control input data from the trainee, and forcibly executing deceleration and hovering operations or attitude correction operations, thereby effectively preventing the virtual drone from colliding or going out of control. This proactive intervention mechanism can improve the safety of the training process, reduce accidents caused by trainees' operational errors, and allow trainees to focus more on learning and improving their flight skills. Simultaneously, the mechanism of high-priority control logic output overriding low-priority control logic output ensures that the most reasonable and safest decision can be made when multiple risks coexist, providing trainees with a more stable, reliable, and efficient training environment.

[0051] In one embodiment, the virtual simulation platform includes a physical simulation module, which is used for: Drive the virtual drone to perform corresponding flight maneuvers based on flight control commands; Based on a preset flight physics model, the virtual drone's attitude angle data, velocity data, and motion angular velocity data are calculated in real time. Based on a preset aerodynamic model, the wind resistance parameters experienced by the virtual drone are calculated in real time. The attitude angle data, velocity data, motion angular velocity data, and wind field drag parameters are fed back to the data conversion module.

[0052] In this embodiment, the physics simulation module is a component of the virtual simulation platform used to simulate the realistic physical behavior of a virtual drone in a virtual environment. The physics simulation module can be an independent software unit or a subsystem integrated within the virtual simulation platform, such as through the physics engine functionality of a game engine. Its function is to calculate the motion state and force conditions of the virtual drone based on received control commands and environmental parameters, ensuring the realism of the simulation. Driving the virtual drone to execute corresponding flight actions according to flight control commands means that the physics simulation module receives flight control commands from the data conversion module and converts the control signals into specific movements of the virtual drone in the virtual world. For example, flight control commands may include parameters such as throttle, pitch, roll, and yaw. The physics simulation module calculates the drone's position, attitude, and velocity at the next time step based on these parameters and the virtual drone's own physical properties. The flight physics model is a mathematical model describing the drone's motion in the air. It typically considers multiple factors such as the drone's mass, inertia, rotor thrust, and drag. Real-time calculation ensures that these kinematic and dynamic data can be updated at a high frequency, thereby guaranteeing the smoothness and accuracy of the simulation. Aerodynamic models are used to simulate the effects of air on drone flight, including air resistance, lift, and lateral forces. Among them, wind field drag parameters are specific quantitative values ​​of these aerodynamic effects, such as wind speed, wind direction, and turbulence intensity. Real-time calculations can ensure that the forces acting on the virtual drone under different flight states and environmental conditions can be accurately reflected.

[0053] For example, the virtual simulation platform can use mainstream game engines, such as Unity 3D or Unreal Engine, to build its physics simulation module. In the physics simulation module, the virtual drone can be modeled as an object with six degrees of freedom rigid body dynamics. The flight physics model can be implemented based on the Newton-Euler equations, where flight control commands (such as throttle, pitch, roll, and yaw inputs) are converted into thrust and torque acting on the drone, and then the drone's position, attitude, linear velocity, and angular velocity are updated in real time using numerical integration methods (such as the Runge-Kutta method). For example, the throttle command can be mapped to the total thrust of the rotor, while roll, pitch, and yaw commands generate corresponding torques by changing the speed difference between different rotors. The aerodynamic model can be integrated into the physics engine or implemented as a separate script, which can calculate the air resistance acting on the drone's fuselage and rotors based on the virtual drone's current speed, attitude, and parameters such as wind speed and direction in the virtual environment. For example, drag can be simplified to a force proportional to the square of the velocity, taking into account drag coefficients in different directions. Finally, the real-time calculated attitude angle data (such as Euler angles or quaternions), velocity data (such as three-dimensional linear velocity vectors), and motion angular velocity data (such as three-dimensional angular velocity vectors), as well as wind field drag parameters, are sent to the data conversion module at a high frequency (e.g., 60 frames per second) through a specific API interface or data bus.

[0054] Through the aforementioned technical solutions, the virtual simulation platform can provide highly realistic drone flight simulations. Specifically, the physics simulation module, based on preset flight physics and aerodynamic models, calculates the virtual drone's attitude angles, velocity, angular velocity, and wind resistance parameters in real time. This ensures that the virtual drone's flight maneuvers and force conditions accurately reflect the physical characteristics and environmental interactions of a real drone, providing accurate input for the data conversion module to generate force feedback control signals. Based on this, the damping force felt by trainees through the dedicated controller will more closely resemble the feel of a real drone under different flight states and environmental conditions, enhancing the immersion and effectiveness of training. This precise physical simulation helps trainees establish correct flight perception and operating habits, effectively compensating for the shortcomings of traditional simulators in terms of physical realism and providing a solid technical foundation for drone VR control training.

[0055] In one embodiment, the virtual simulation platform includes an obstacle detection module, which is used for: Using the virtual drone as the origin, project several collision detection rays along a preset direction; When the collision detection beam hits an obstacle, the coordinates of the hit point and the obstacle's label are obtained; The distance between the virtual drone and the obstacle is calculated based on the coordinates of the hit point, obstacle distance data is generated, and the obstacle distance data is sent to the data conversion module.

[0056] In this embodiment, the obstacle detection module is a functional unit within the virtual simulation platform. Its main function is to monitor the spatial relationship between the virtual drone and obstacles in the virtual environment in real time and identify potential collision risks. The obstacle detection module can exist as an independent component or integrated service of the virtual simulation platform, responsible for handling logic and data related to obstacle detection. It can be implemented through software algorithm modules, such as integrating a collision detection API into a game engine. The obstacle detection module projects several collision detection rays along preset directions from the virtual drone as the origin. This means the system emits virtual rays from the current position of the virtual drone in multiple directions around it. These collision detection rays are virtual detection tools used to detect the presence of obstacles in the virtual environment. The preset directions can include multiple key directions such as the front, rear, left, right, above, below, and diagonal directions of the drone, achieving omnidirectional obstacle perception. Furthermore, the projection of the rays can be implemented through a ray projection algorithm or through proximity detection using a voxel-based or mesh-based spatial partitioning method. When a collision detection ray hits an obstacle, the obstacle detection module acquires the coordinates of the hit point and the obstacle's tag. The hit point coordinates refer to the precise three-dimensional position where the ray intersects the obstacle's surface, while the obstacle tag identifies the type or attribute of the hit obstacle, such as a building, tree, or other aircraft. The hit point coordinates are obtained using the intersection information returned by the ray casting algorithm, while the obstacle tag can be pre-stored in the obstacle model data of the virtual scene and retrieved upon hit. After acquiring the hit point coordinates, the obstacle detection module further calculates the Euclidean distance or other spatial distance between the hit point and the virtual drone's current position, thus obtaining the distance between the virtual drone and the nearest obstacle. Distance calculation can be achieved using distance formulas in three-dimensional space, such as the straight-line distance between two points. After generating the obstacle distance data, the obstacle detection module transmits it as crucial environmental perception information to the data conversion module. The data conversion module can use this distance data to adjust the force feedback control signal generation logic, such as increasing damping force or triggering a warning when approaching an obstacle. Furthermore, data transmission can be achieved through internal message queues, shared memory, or function calls, ensuring that data can be transmitted in real time and accurately.

[0057] Furthermore, the obstacle detection module is also used to: determine the type of obstacle based on the obstacle label, including static obstacles and dynamic obstacles; When the obstacle type is a static obstacle, a collision warning is issued using a preset first detection distance threshold; When the obstacle type is dynamic obstacle, a collision warning is issued using a preset second detection distance threshold, wherein the second detection distance threshold is greater than the first detection distance threshold.

[0058] Furthermore, the obstacle detection module is also used for: Based on obstacle distance data and virtual drone speed data, calculate the collision time between the virtual drone and the obstacle; When the collision time is less than the preset collision time threshold, a collision warning signal is generated and sent to the data conversion module.

[0059] As an example, the collision time threshold can be set to 1.5 seconds. Based on the TTC algorithm, the collision time between the virtual drone and the obstacle is calculated in real time. When the collision time is less than 1.5 seconds, a collision warning signal is generated to prompt the trainee to take immediate evasive action.

[0060] For example, the obstacle detection module in a virtual simulation platform can utilize the physics engine functionality provided by mainstream game engines (such as Unity 3D or Unreal Engine). For instance, in Unity 3D, the obstacle detection module can periodically (e.g., every frame or at fixed time intervals) call the `Physics.RaycastAll` function from the virtual drone's position, specifying multiple preset direction vectors as the projection direction of the light rays. These direction vectors can be predefined as local directions in the drone's body coordinate system, such as `transform.forward`, `transform.right`, `transform.up` and their negative directions, as well as some diagonal directions, to cover the perception range around the drone. When the `RaycastAll` function returns one or more `RaycastHit` objects, it indicates that a ray has hit an obstacle. The obstacle detection module can extract the `point` property from each `RaycastHit` object as the coordinates of the hit point and obtain the obstacle label from the `collider.gameObject.layer` property, such as "Building", "Tree", "DynamicObstacle", etc. Furthermore, the obstacle detection module can calculate the Euclidean distance between the virtual drone's current position and `RaycastHit.point`, thereby generating obstacle distance data. For example, Vector3.Distance can be used for calculation. Finally, the obstacle detection module encapsulates the obstacle distance data into a data packet and sends it to the data conversion module in real time via the message bus or event system within the virtual simulation platform. For example, an event containing obstacle distance data can be triggered, which the data conversion module subscribes to and receives. After receiving this data, the data conversion module adjusts the generation of the force feedback control signal according to preset logic, thereby generating a corresponding damping force on the dedicated controller to alert the trainee to avoid obstacles.

[0061] Through the aforementioned technical solution, the system of this application can perceive the distance between the virtual drone and obstacles in the environment in real time, and dynamically adjust the force feedback according to potential collision risks. This solves the problem of lack of environmental perception and obstacle avoidance feedback in traditional VR training, enabling trainees to receive timely prompts of approaching obstacles during training, and to intuitively feel the physical feedback of operational limitations or the need for obstacle avoidance through the damping force generated by the dedicated controller. This mechanism not only enhances the realism and immersion of the training but also effectively cultivates trainees' ability to perform drone obstacle avoidance flight in complex environments, reducing the risk of collisions during actual flight, thereby improving the teaching effectiveness and safety of the training system.

[0062] In one embodiment, the control input data includes throttle control data, yaw control data, pitch control data, roll control data, and flight mode switching data. The dedicated controller is a 1:1 replica of a real drone remote controller, and the interaction unit includes: The left joystick is used to collect throttle control data and yaw control data. The right joystick is used to collect pitch control data and roll control data. A mode switching switch, used to collect flight mode switching data; The force feedback unit includes force feedback motors integrated inside the left and right joysticks. The force feedback motors are used to generate corresponding damping forces based on the force feedback control signal.

[0063] In this embodiment, the control input data refers to the command signals generated when the trainee operates the virtual drone through a dedicated controller, serving as a digital representation of the trainee's operational intentions. The control input data includes throttle control data, yaw control data, pitch control data, roll control data, and flight mode switching data. Specifically, throttle control data controls the vertical thrust of the virtual drone, affecting its ascent or descent speed; yaw control data controls the rotation of the virtual drone in the horizontal plane, affecting its nose direction; pitch control data controls the upward or downward tilt of the virtual drone's nose, affecting its forward or backward speed; roll control data controls the left or right tilt of the virtual drone's body, affecting its left or right translation speed; and flight mode switching data is used to switch between different flight modes to adapt to different training needs or flight scenarios. The aforementioned control input data is typically sent from the dedicated controller in the form of analog or digital signals. For example, the physical position of a joystick can be converted into an analog voltage signal, which is then converted into digital data by an analog-to-digital converter; the on / off state of a switch directly generates a digital signal. The dedicated controller is a 1:1 replica of a real drone remote controller, meaning it maintains a high degree of consistency with the actual drone remote controller in terms of appearance, feel, button layout, joystick travel, and damping. This is achieved through 3D scanning, physical measurement, and material analysis of the real drone remote controller, followed by high-precision mold manufacturing and customized electronic component integration. For example, the joystick's spring force, friction, and centering accuracy should all match those of the real remote controller. The left joystick is typically used to control throttle and yaw, and its implementation can include Hall sensors, potentiometers, or optical encoders to accurately measure the joystick's positional changes in two dimensions. The right joystick is typically used to control pitch and roll, and its implementation is similar to the left joystick, also using Hall sensors, potentiometers, or optical encoders. The mode switch is used to select different flight modes, and its implementation can include toggle switches, push-button switches, or rotary encoders to provide clear mode switching feedback. A force feedback motor is an actuator that generates precise torque or force based on electrical signals. By directly mounting the force feedback motor within the mechanical structure of the left and right joysticks, it can directly influence the joystick's movement, thereby altering the joystick's damping or generating vibrations. This can be achieved, for example, using a DC motor, stepper motor, or voice coil motor, and then employing reduction gears or linkage mechanisms to transmit the motor's force to the joystick. Another approach is to use an electromagnetic brake, adjusting the joystick's damping by changing the strength of the electromagnetic field.

[0064] For example, the dedicated controller can be constructed with a shell made of high-strength engineering plastics or metal alloys, with dimensions, weight, and grip consistent with mainstream consumer-grade drone remote controllers on the market. Both the left and right joysticks can employ high-precision three-axis Hall sensors to ensure accurate acquisition of throttle, yaw, pitch, and roll operations. Their joystick travel and return-to-center force are precisely calibrated to simulate the feel of a real remote controller. The mode switch can be a three-position toggle switch for switching between "attitude mode," "GPS mode," and "manual mode." The force feedback unit can integrate a miniature brushless DC motor within the base of each joystick. This miniature brushless DC motor is connected to the joystick's shaft via a precision gear set. When the data conversion module sends a force feedback control signal, such as simulating air resistance encountered by a virtual drone during high-speed flight, the motor generates a torque opposite to the joystick's movement direction, thereby increasing the joystick's damping feel. When a virtual drone is suddenly affected by a gust of wind in the air, causing its attitude to deviate, the force feedback motor can generate an instantaneous pulse force to simulate the feeling of the joystick being pushed by an external force, prompting the trainee to make corrections.

[0065] The aforementioned technical solution designs a dedicated controller that is a 1:1 replica of a real drone remote controller, refining the types of input data. Force feedback motors are integrated into the left and right joysticks of the interactive unit, allowing trainees to experience physical and tactile feedback highly consistent with real drone operation during VR control training. This high-fidelity control experience effectively addresses the shortcomings of traditional simulators in terms of realism, preventing trainees from developing muscle memory in virtual training that is inconsistent with real operation. Through precise throttle, yaw, pitch, roll control data, and mode switching data, trainees can more intuitively control the virtual drone. The damping force generated by the force feedback motors allows them to perceive the external environmental influences and changes in their flight status in real time, thereby enhancing the immersion, realism, and effectiveness of the training.

[0066] In one embodiment, the virtual simulation platform includes an evaluation module and an interactive display module, the evaluation module being used for: Real-time acquisition of 3D coordinate data, flight altitude data, attitude angle data, and flight heading data of virtual drones; Retrieve a preset assessment standard library, which includes standard route coordinates, fixed target positions, specified attitude angles, standard headings, and allowable error thresholds; Based on the differences between the three-dimensional coordinate data and the standard route coordinates, the differences between the attitude angle data and the specified attitude angles, and the differences between the flight heading data and the standard heading, the position offset, attitude tilt error, and heading deviation are calculated. Wherein, when the operation mode information is training mode information, the position offset, attitude inclination error and heading deviation are sent to the interactive display module for real-time display; When the operation mode information is assessment mode information, the position offset, attitude inclination error and heading deviation are recorded, and an assessment report is generated and sent to the trainee terminal.

[0067] In this embodiment, the assessment module is a core functional unit for performing quantitative analysis and evaluation on the flight performance of the virtual drone. It can be used as an independent software service module inside the virtual simulation platform to perform data interaction with other components of the platform through a clearly defined interface, or be integrated and implemented as part of the core logic of the platform. The interactive display module is configured to present the output result of the assessment module to the trainee in a visualized manner, which can be implemented through a head-up display interface superimposed in a VR head-mounted display, or perform information display through an external display connected to the trainee terminal. The three-dimensional coordinate data, flight altitude data, attitude angle data and flight heading data of the virtual drone are collected in real time. These data can be provided in real time by the physics engine or rendering engine of the virtual simulation platform, and obtained through a data stream or event subscription mechanism. The preset assessment standard library is a benchmark for measuring the flight performance of the trainee, the content of which can be stored in a local database of the virtual simulation platform, or exist in the form of configuration files (such as XML, JSON), and can be flexibly configured according to different training subjects and difficulty levels. The position offset, attitude inclination error and heading deviation are calculated according to the difference between each piece of flight data and the assessment standard, which can be implemented by mathematical algorithms such as Euclidean distance calculation and angle difference calculation. In the training mode, the error data is sent to the interactive display module in real time for display, aiming to provide immediate feedback and help the trainee discover and correct operation errors in time. In the assessment mode, recording the error data and generating the assessment report focuses on the summative evaluation of the overall performance of the trainee, so as to provide a basis for subsequent skill analysis and improvement.

[0068] Further, as an example, the allowable error of the position offset may be set to ±0.5 meters. In training subjects such as figure-eight flight and fixed-point hovering, the deviation between the actual position of the virtual drone and the coordinates of the standard route is calculated in real time. When the position offset is within the range of ±0.5 meters, it is determined to be qualified; when the position offset exceeds ±0.5 meters, it is determined to be unqualified and corresponding points are deducted.

[0069] The allowable error of the attitude inclination error may be set to ±5°. In training subjects such as horizontal flight and fixed-point hovering, the deviation between the actual attitude angle of the virtual drone and the specified attitude angle is calculated in real time. When the attitude inclination error is within the range of ±5°, it is determined to be qualified; when the attitude inclination error exceeds ±5°, it is determined to be unqualified and corresponding points are deducted.

[0070] The allowable heading deviation error may be set to ±10°. In training subjects such as route flight and target tracking, the deviation between the actual heading of the virtual unmanned aerial vehicle and the standard heading is calculated in real time. When the heading deviation is within the range of ±10°, it is determined to be qualified; when the heading deviation exceeds ±10°, it is determined to be unqualified and corresponding points are deducted.

[0071] Exemplarily, the virtual simulation platform may be configured with an assessment module. In terms of software architecture, the assessment module may be an independent subsystem, which performs data interaction with the physics engine and the rendering engine of the virtual unmanned aerial vehicle through an application programming interface. For example, the physics engine may output attitude angle data, velocity data, motion angular velocity data, three-dimensional coordinate data, flight altitude data and flight heading data of the virtual unmanned aerial vehicle at a frequency of 60 frames per second, and the assessment module subscribes to these data streams in real time. The assessment standard library may be a JSON configuration file stored in the resource directory of the virtual simulation platform, in which standard parameters of different training subjects are defined, such as "fixed-point hovering", "figure-eight circling flight" and "route patrol". For example, for the "fixed-point hovering" subject, the assessment standard library may include target three-dimensional coordinates, an allowable position offset threshold and an allowable attitude inclination error threshold. When a trainee selects the "fixed-point hovering" training mode, the assessment module will calculate the Euclidean distance between the current three-dimensional coordinates of the virtual unmanned aerial vehicle and the target three-dimensional coordinates in real time as the position offset, and calculate the difference between the current attitude angle and the horizontal attitude as the attitude inclination error, and these calculation results will be immediately sent to the interactive display module through the internal message bus. The interactive display module may display information such as "Position offset: 0.3 meters (green)" or "Attitude inclination error: 5 degrees (red)" in real time in the upper left corner of the screen in the form of a semi-transparent head-up display in the trainee's VR head-mounted display, and may be accompanied by a direction indicator arrow to guide the trainee to make corrections. When the trainee enters the "assessment mode" to carry out the "route patrol" assessment, the assessment module also collects the aforementioned flight data in real time and compares it with the standard route coordinate sequence preset in the assessment standard library, so as to calculate the position offset along the route and the heading deviation. However, at this time, such error information will not be displayed to the trainee in real time; instead, the assessment module will record the error data together with the timestamp in a log file. After the assessment is finished, the assessment module will process the log data to generate an assessment report, which includes the trainee's flight trajectory diagram, curves of various errors changing with time, as well as the final assessment score and the details of point deduction items. Further, this report may then be sent to the personal training file at the trainee end.

[0072] Herein, an example diagram of the assessment report can be seen in Figure 2 .

[0073] Through the above technical solutions, the UAV VR control training system of this application can provide quantitative evaluation and targeted feedback on trainees' flight performance. Specifically, in training mode, trainees can obtain real-time deviation information between their own operations and standards, such as position offset, attitude tilt error, and heading deviation, thereby enabling them to adjust control strategies immediately, effectively correct errors, and improve training efficiency and skill mastery speed. In assessment mode, the system can objectively record trainees' flight data and generate detailed evaluation reports, providing trainees with comprehensive assessment results and improvement suggestions, reducing the subjectivity and inconsistency of human evaluation, and ensuring the fairness and accuracy of the assessment. By introducing this evaluation and feedback mechanism, the entire training process can be more scientific and efficient, enhancing the trainees' learning experience and training effectiveness.

[0074] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0075] In one embodiment, a VR control training method for unmanned aerial vehicles (UAVs) driven by a dedicated simulation controller is provided. This method corresponds to the VR control training system for UAVs driven by a dedicated simulation controller described in the previous embodiment. The VR control training method for UAVs driven by a dedicated simulation controller includes the following steps: The student inputs control input data through the interaction unit of the dedicated controller, and the interaction unit sends the control input data to the data conversion module; The data conversion module generates flight control commands based on the control input data and sends the flight control commands to the virtual simulation platform; The virtual simulation platform receives flight control commands, drives the virtual drone to perform corresponding flight actions, and acquires the corresponding flight data set; The virtual simulation platform feeds back flight data sets and operational mode information to the data conversion module; The flight data set includes attitude angle data, velocity data, and motion angular velocity data; the operation mode information includes training mode information and assessment mode information; the data conversion module generates a force feedback control signal based on the flight data set and the current operation mode information, and sends the force feedback control signal to a dedicated controller, specifically including: When the running mode information is training mode information, the eigencomponents are generated based on the attitude angle data, velocity data and motion angular velocity data, and the guidance components are generated based on the deviation between the attitude angle data and the preset reference attitude angle and the deviation between the motion angular velocity data and the preset reference angular velocity. When the virtual drone is determined to be in normal flight state, the intrinsic component and the guidance component are superimposed according to the preset first weight ratio to generate a force feedback control signal. When the virtual drone is determined to be in a dangerous flight state, the intrinsic component and the guidance component are superimposed according to the preset second weight ratio to generate a force feedback control signal. When the operating mode information is the assessment mode information, a force feedback control signal containing only intrinsic components is generated based on the flight data set. The force feedback unit of the dedicated controller generates a corresponding damping force based on the force feedback control signal, enabling the learner to perceive the flight status of the virtual drone through the damping force.

[0076] For specific limitations regarding a VR control training method for UAVs driven by a dedicated simulation controller, please refer to the limitations of a VR control training system for UAVs driven by a dedicated simulation controller mentioned above, which will not be repeated here.

[0077] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A VR control training system for unmanned aerial vehicles driven by a simulation-specific controller, comprising a student terminal, characterized in that, It also includes a dedicated controller, a data conversion module, and a virtual simulation platform; The dedicated controller includes a force feedback unit and an interaction unit. The interaction unit is used to collect the control input data from the student's end and send it to the data conversion module. The force feedback unit is used to generate a corresponding damping force based on the force feedback control signal sent by the data conversion module. The data conversion module is used to generate flight control commands based on the control input data and send them to the virtual simulation platform; it is also used to receive the flight data set and operating mode information fed back by the virtual simulation platform, and generate force feedback control signals based on the flight data set and the current operating mode information. The virtual simulation platform is used to acquire mode configuration information and generate operating mode information based on the mode configuration information; it is also used to receive flight control commands, drive the virtual drone to perform corresponding flight actions, and acquire the corresponding flight data set. The flight dataset includes attitude angle data, velocity data, and motion angular velocity data; the operation mode information includes training mode information and assessment mode information; the data conversion module includes: The first conversion unit, when the operating mode information is training mode information, is used to calculate and generate eigencomponents based on attitude angle data, velocity data, and motion angular velocity data, and to calculate and generate guidance components based on the deviations between the attitude angle data and preset reference attitude angles and the deviations between the motion angular velocity data and preset reference angular velocities; wherein, the first conversion unit is further used to: When the virtual drone is determined to be in normal flight state, the intrinsic component and the guidance component are superimposed according to the preset first weight ratio to generate a force feedback control signal. When the virtual drone is determined to be in a dangerous flight state, the intrinsic component and the guidance component are superimposed according to the preset second weight ratio to generate a force feedback control signal. The second conversion unit, when the operating mode information is the assessment mode information, is used to generate a force feedback control signal that includes only intrinsic components based on the flight data set.

2. The VR control training system for unmanned aerial vehicles driven by a simulation-specific controller according to claim 1, characterized in that: The first conversion unit is specifically used for: Real-time acquisition of obstacle distance data of virtual drones, and retrieval of preset reference attitude angles and reference angular velocities; The wind resistance parameters experienced by the virtual drone are obtained, and the eigencomponents are generated based on the attitude angle data, velocity data, motion angular velocity data and wind resistance parameters through a preset aerodynamic mapping model. Based on the deviation between the attitude angle data and the reference attitude angle, and the deviation between the motion angular velocity data and the reference angular velocity, the guide component is calculated and generated through the preset PID correction mapping model. When the obstacle distance data is greater than the preset safe distance threshold and the attitude angle data does not exceed the preset first tilt angle threshold, it is determined to be a normal flight state. When the obstacle distance data is not greater than the preset safe distance threshold, or the attitude angle data exceeds the preset first tilt angle threshold, it is determined to be a dangerous flight state.

3. The VR control training system for unmanned aerial vehicles driven by a simulation-specific controller according to claim 1, characterized in that: The second conversion unit is specifically used for: Real-time acquisition of wind resistance parameters experienced by the virtual drone; Based on attitude angle data, velocity data, motion angular velocity data, and wind field drag parameters, eigencomponents are generated through a pre-set aerodynamic mapping model. The intrinsic components are directly output as force feedback control signals, and flight control commands are generated based on the control input data from the student's end.

4. The VR control training system for unmanned aerial vehicles driven by a simulation-specific controller according to claim 2, characterized in that: The data conversion module further includes a control priority unit, which is used for: A control priority sequence is set, which includes emergency collision avoidance logic, fuselage extreme attitude protection, flight physical parameter limiting, auxiliary stabilization algorithm, and control input data processing in sequence. When the distance data of the detected obstacle is less than the preset emergency distance threshold, the emergency collision avoidance logic is triggered, locking the control input data in the direction of the obstacle and forcibly performing deceleration and hovering operation; When the attitude angle data exceeds the preset second tilt angle threshold, or the motion angular velocity data exceeds the preset angular velocity threshold, the fuselage extreme attitude protection is triggered, the control input data of the student terminal is cut off and the attitude return operation is forcibly executed. When several high-priority control logics are triggered simultaneously, they are executed sequentially according to the control priority sequence. The output of the high-priority control logic will override the output of the low-priority control logic. When the virtual drone is in normal flight, the intrinsic components and the guidance components are mixed according to the first weight ratio, and flight control commands are generated by combining the control input data.

5. A VR control training system for unmanned aerial vehicles driven by a simulation-specific controller according to any one of claims 1-4, characterized in that: The virtual simulation platform includes a physical simulation module, which is used for: Drive the virtual drone to perform corresponding flight maneuvers based on flight control commands; Based on a preset flight physics model, the virtual drone's attitude angle data, velocity data, and motion angular velocity data are calculated in real time. Based on a preset aerodynamic model, the wind resistance parameters experienced by the virtual drone are calculated in real time. The attitude angle data, velocity data, motion angular velocity data, and wind field drag parameters are fed back to the data conversion module.

6. A VR control training system for unmanned aerial vehicles driven by a simulation-specific controller according to any one of claims 1-4, characterized in that: The virtual simulation platform includes an obstacle detection module, which is used for: Using the virtual drone as the origin, project several collision detection rays along a preset direction; When the collision detection beam hits an obstacle, the coordinates of the hit point and the obstacle's label are obtained; The distance between the virtual drone and the obstacle is calculated based on the coordinates of the hit point, obstacle distance data is generated, and the obstacle distance data is sent to the data conversion module.

7. The VR control training system for unmanned aerial vehicles driven by a simulation-specific controller according to claim 1, characterized in that: The control input data includes throttle control data, yaw control data, pitch control data, roll control data, and flight mode switching data. The dedicated controller is a 1:1 replica of a real drone remote controller. The interaction unit includes: The left joystick is used to collect throttle control data and yaw control data. The right joystick is used to collect pitch control data and roll control data. A mode switching switch, used to collect flight mode switching data; The force feedback unit includes force feedback motors integrated inside the left and right joysticks. The force feedback motors are used to generate corresponding damping forces based on the force feedback control signal.

8. The VR control training system for unmanned aerial vehicles driven by a simulation-specific controller according to claim 1, characterized in that: The virtual simulation platform includes an evaluation module and an interactive display module. The evaluation module is used for: Real-time acquisition of 3D coordinate data, flight altitude data, attitude angle data, and flight heading data of virtual drones; Retrieve a preset assessment standard library, which includes standard route coordinates, fixed target positions, specified attitude angles, standard headings, and allowable error thresholds; Based on the differences between the three-dimensional coordinate data and the standard route coordinates, the differences between the attitude angle data and the specified attitude angles, and the differences between the flight heading data and the standard heading, the position offset, attitude tilt error, and heading deviation are calculated. When the operating mode information is the training mode information, the position offset, attitude tilt error and heading deviation are sent to the interactive display module for real-time display. When the operating mode information is the assessment mode information, the position offset, attitude tilt error and heading deviation are recorded, and an assessment report is generated and sent to the trainee's terminal.

9. A VR control training method for unmanned aerial vehicles driven by a simulation-specific controller, characterized in that, Including the following steps: The student inputs control input data through the interaction unit of the dedicated controller, and the interaction unit sends the control input data to the data conversion module; The data conversion module generates flight control commands based on the control input data and sends the flight control commands to the virtual simulation platform; The virtual simulation platform receives flight control commands, drives the virtual drone to perform corresponding flight actions, and acquires the corresponding flight data set; The virtual simulation platform feeds back flight data sets and operational mode information to the data conversion module; The flight data set includes attitude angle data, velocity data, and motion angular velocity data; the operation mode information includes training mode information and assessment mode information; the data conversion module generates a force feedback control signal based on the flight data set and the current operation mode information, and sends the force feedback control signal to a dedicated controller, specifically including: When the running mode information is training mode information, the eigencomponents are calculated based on the attitude angle data, velocity data and motion angular velocity data, and the guidance components are calculated based on the deviation between the attitude angle data and the preset reference attitude angle and the deviation between the motion angular velocity data and the preset reference angular velocity. When the virtual drone is determined to be in normal flight state, the intrinsic component and the guidance component are superimposed according to the preset first weight ratio to generate a force feedback control signal. When the virtual drone is determined to be in a dangerous flight state, the intrinsic component and the guidance component are superimposed according to the preset second weight ratio to generate a force feedback control signal. When the operating mode information is the assessment mode information, a force feedback control signal containing only intrinsic components is generated based on the flight data set. The force feedback unit of the dedicated controller generates a corresponding damping force based on the force feedback control signal, enabling the learner to perceive the flight status of the virtual drone through the damping force.

Images