An unmanned aerial vehicle multi-sensor fusion-oriented simulation test verification platform

Abstract

The application relates to a simulation test verification platform for unmanned aerial vehicle multi-sensor fusion, which comprises a scene simulation module, a flight vehicle simulation module, a sensor simulation module and a flight control module; the scene simulation module is used for providing a simulation scene and meteorological and scenario settings; the flight vehicle simulation module is used for adding an unmanned aerial vehicle simulation model to the simulation scene; the sensor simulation module is used for adding required sensors to the unmanned aerial vehicle simulation model and grabbing sensor data; and the flight control module is used for generating a flight route of the unmanned aerial vehicle and controlling the movement of the unmanned aerial vehicle, wherein the control comprises position control and speed control. The application can synchronously acquire different sensor data in a simulation scene in real time, has different sensor position conversion matrices, meets the requirements of multi-sensor fusion algorithms on time synchronization and space synchronization of different sensor data, reduces the cost of real machine test verification, and guarantees the stability and reliability of unmanned aerial vehicle intelligent functions based on multi-sensor fusion.

Description

Technical Field

[0001] This invention relates to the field of simulation testing technology, specifically a simulation testing and verification platform for multi-sensor fusion in unmanned aerial vehicles (UAVs). Background Technology

[0002] In recent years, drones, as a new type of aircraft, have been widely used in various fields, which has also continuously raised higher requirements for their environmental perception capabilities and intelligence levels. Equipping drones with different sensors and fusing multi-sensor data (also known as multi-sensor fusion) to develop new intelligent functions has become a research hotspot in the drone field. At the same time, with the development of computer vision, visual sensors such as cameras are being used extensively on drones, and the fusion of data from other sensors with camera data is constantly generating new insights. For example, the emergency landing function for drones developed based on the fusion of image data and point cloud data, through onboard cameras and LiDAR, enriches the perception of the surrounding environment in safety warning situations such as communication link loss or insufficient power preventing reaching the alternative landing point. This enables proactive identification of landable areas in unknown environments and safe landing of the drone.

[0003] Because the algorithms and intelligent functions of multi-sensor fusion in UAVs are quite complex, and the application scenarios vary greatly, the model parameters in the algorithms require multiple iterations of debugging, and the sensor installation requires multiple adjustments and configurations. Real-world testing carries high uncertainty and risk. Therefore, a simulation testing and verification platform is urgently needed, providing strong scene rendering and sensor simulation capabilities.

[0004] Currently, mainstream drone simulation platforms include Gazebo simulation system and Airsim simulation system.

[0005] The Gazebo simulation system offers high-fidelity physical simulations, a complete set of sensor models, and a relatively user-friendly interface. It is typically used in conjunction with ROS (Robot Operating System, providing data input / output communication links) to provide developers with a basic simulation environment. However, Gazebo cannot provide realistic simulation scenarios and cannot meet the extremely high image data requirements of cameras. Furthermore, it can only collect sensor data and cannot perform end-to-end testing and verification simulations, including model inference, multi-sensor fusion algorithms, and intelligent UAV function outputs.

[0006] Airsim is a robot simulation system developed by Microsoft based on Unreal Engine. It supports physical and visual simulation of robots such as drones and unmanned vehicles, and is suitable for deep learning-based visual AI simulation verification. Airsim provides many API interfaces for data reading, vehicle control, weather control, and more. While Airsim has strong scene rendering capabilities, its sensor data clocks are inconsistent, failing to meet the requirements for time-synchronized sensor data fusion. Furthermore, it can only collect sensor data and cannot perform end-to-end testing and verification simulations, including model inference, multi-sensor fusion algorithms, and intelligent drone function output.

[0007] The information disclosed in this background section is intended only to enhance the understanding of the general background of the invention and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0008] To address the shortcomings of existing technologies, the present invention aims to provide a simulation testing and verification platform for multi-sensor fusion in unmanned aerial vehicles (UAVs). This platform solves the problems of insufficient scene rendering capabilities, inability to support time and space synchronization of multi-sensor data, lack of end-to-end simulation of multi-sensor fusion algorithms and intelligent functions of UAVs, and high cost and risk of real-machine testing and verification of multi-sensor fusion perception and intelligent functions of UAVs.

[0009] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0010] A simulation test and verification platform for multi-sensor fusion of unmanned aerial vehicles (UAVs) is characterized by comprising: a scene simulation module, an aircraft simulation module, a sensor simulation module, and a flight control module.

[0011] The scene simulation module uses Unreal Engine to generate .uproject files for storing simulation scenes, providing three-dimensional virtual drone simulation scenes, and providing weather and scenario settings within the simulation scenes;

[0012] The aircraft simulation module is used to add UAV simulation models to the simulation scene, to provide UAV simulation models that conform to the kinematics and dynamics characteristics of the real aircraft, and to set the UAV shape and flight control through the UAV simulation model;

[0013] The sensor simulation module is used to add the required sensors to the UAV simulation model and capture sensor data in the simulation scene.

[0014] The sensor simulation module supports time and space synchronization of data acquisition from different sensors by setting a unified clock and a relative position transformation matrix for different sensors. The relative position transformation matrix for different sensors is directly input into the multi-sensor fusion algorithm for processing.

[0015] The flight control module is used to generate the flight path of the UAV and to control the movement of the UAV. The control includes position control and speed control.

[0016] Based on the above technical solution, the simulation scene includes several typical terrain scenes, specifically divided into: basic terrain scene, basic feature scene, and basic natural scene.

[0017] The meteorological and scenario settings include: meteorological conditions such as rain, snow, dust storms, fog, fallen leaves, road surface humidity, road surface snow, and fallen leaves; and scenarios such as: lighting conditions, rain and snow, dust storms, fog and haze, and natural wind.

[0018] Based on the above technical solution, the sensor model includes: camera, lidar, GPS, IMU, barometer, magnetometer, and distance sensor;

[0019] The sensor model supports personalized settings for the sensor, including installation location, installation angle, and sensor coefficients.

[0020] Based on the above technical solutions, it also includes: a communication module, a model inference module, a multi-sensor fusion algorithm module, and an intelligent function output module;

[0021] The communication module is used to transmit sensor data and flight control commands through the ROS publish and subscribe mechanism, wherein: the sensor data is transmitted from the sensor simulation module to the model inference module, and the control commands refer to flight control commands, which are transmitted from the intelligent function output module to the flight control module; it is also used to set the transmission frequency of sensor data to meet the real-time requirements of simulation.

[0022] The model inference module is used to iterate a high-performance model based on sensor data, and to process the sensor data using the high-performance model.

[0023] The multi-sensor fusion algorithm module fuses data from different sensors based on the data processing results output by the model inference module.

[0024] The intelligent function output module acquires the data fusion results output by the multi-sensor fusion algorithm module, outputs UAV control commands, and realizes intelligent functions.

[0025] Based on the above technical solution, the multi-sensor fusion algorithm module is part of the simulation test and verification platform. It takes the processing result of the model inference module as input and then fuses the data from different sensors.

[0026] Based on the data fusion results from the multi-sensor fusion algorithm module, the intelligent function output module sends UAV control commands to the flight control module to achieve the corresponding intelligent functions.

[0027] A simulation testing and verification method for multi-sensor fusion in unmanned aerial vehicles (UAVs), characterized by applying any of the aforementioned simulation testing and verification platforms, the method comprising the following steps:

[0028] The aircraft parameters are pre-set in the simulation test and verification platform according to the simulation requirements;

[0029] Sensor parameters are pre-set in the simulation test and verification platform according to simulation requirements;

[0030] Flight route parameters are pre-set in the simulation test and verification platform according to simulation requirements;

[0031] The scene simulation module generates a three-dimensional virtual simulation scene, the aircraft simulation module adds a drone simulation model to the three-dimensional virtual simulation scene, and the sensor simulation module adds the required sensors to the drone simulation model.

[0032] The flight control module controls the movement of the UAV, which moves along a pre-set flight path in a three-dimensional virtual simulation scene.

[0033] Based on the above technical solution, the following steps are also included:

[0034] The sensor simulation module synchronously captures sensor data in a 3D virtual simulation scene;

[0035] The communication module sends sensor data to the model inference module;

[0036] The model inference module processes the input sensor data using a high-performance model.

[0037] The multi-sensor fusion algorithm module fuses the processed sensor data;

[0038] The intelligent function output module outputs drone control commands based on the results of multi-sensor data fusion, thereby realizing the intelligent functions of the drone.

[0039] Based on the above technical solution, the following steps are also included:

[0040] The communication module transmits the UAV control commands issued by the intelligent function output module to the flight control module, which then controls the aircraft to complete the simulation of the intelligent function.

[0041] The present invention provides a simulation testing and verification platform for multi-sensor fusion in unmanned aerial vehicles (UAVs), which is used to realize the simulation testing and verification of UAV functions based on multi-sensor fusion algorithms, and has the following beneficial effects:

[0042] 1. It can acquire data from different sensors in the simulation scene in real time and has a transformation matrix for different sensor positions, which meets the requirements of multi-sensor fusion algorithm for time synchronization and spatial synchronization of data from different sensors;

[0043] 2. The architecture of the simulation test and verification platform can be applied to different operating scenarios, different sensor fusion algorithms, and different UAV intelligent function testing and verification, and supports end-to-end UAV multi-sensor fusion simulation (end-to-end refers to the closed-loop verification of the entire platform);

[0044] 3. It reduces the cost of real-machine testing and verification, and ensures the stability and reliability of the intelligent functions of drones based on multi-sensor fusion. Attached Figure Description

[0045] The present invention includes the following figures:

[0046] The accompanying drawings are provided to better understand the invention and are not intended to unduly limit the scope of the invention. Wherein:

[0047] Figure 1 The system architecture diagram of Embodiment 1 of the simulation test and verification platform for multi-sensor fusion of unmanned aerial vehicles (UAVs) described in this invention.

[0048] Figure 2 The flowchart is a first embodiment of the simulation test and verification method for multi-sensor fusion of unmanned aerial vehicles (UAVs) described in this invention. Detailed Implementation

[0049] The present invention will be further described in detail below with reference to the accompanying drawings. This detailed description is an illustration in conjunction with exemplary embodiments of the invention, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the invention. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0050] like Figure 1 As shown, this invention provides a simulation test and verification platform for multi-sensor fusion of unmanned aerial vehicles (UAVs), including: a scene simulation module, an aircraft simulation module, a sensor simulation module, and a flight control module;

[0051] The scene simulation module uses Unreal Engine to generate .uproject files for storing simulation scenes, providing three-dimensional virtual drone simulation scenes, and providing weather and scenario settings within the simulation scenes;

[0052] The simulation scenario is a simulation scenario of the intelligent functions of the UAV, such as the simulation scenario of the emergency landing function of the UAV. The verification platform described in this invention can be used for testing and verifying the emergency landing function of the UAV.

[0053] Taking the emergency landing function of drones as an example, the simulation scenario is the urban traffic field. The scenario simulation module provides a set of realistic and large-scale three-dimensional virtual cities as simulation scenarios. The three-dimensional virtual cities conform to the general urban architecture and landforms and have a reasonable layout. They are realistic and beautiful, have good rendering effects, and are stored in the .uproject file.

[0054] The simulation scene includes several typical terrain scenes. Taking the three-dimensional virtual city as an example, it includes the following typical terrain scenes:

[0055] Basic terrain scenes: such as villas, high-rise buildings, factories, commercial buildings, power transmission towers, etc.;

[0056] Basic characteristic scenes: such as cars, pedestrians, streetlights, power lines, highways, paved roads, swimming pools, benches, fences, etc.;

[0057] Basic natural scenes: such as mountains, forests, lakes, grasslands, etc.;

[0058] The simulation scene includes weather and scenario settings. The weather specifically includes: rain, snow, dust storms, fog, fallen leaves, road surface humidity, road surface snow, and fallen leaves on the road surface. The scenarios refer to natural weather conditions, specifically including: lighting conditions, rain and snow weather, dust storms and fog, and natural wind.

[0059] The aircraft simulation module is used to add UAV simulation models to the simulation scene, to provide UAV simulation models that conform to the kinematics and dynamics characteristics of the real aircraft, and to set the UAV shape and flight control through the UAV simulation model;

[0060] Taking the emergency landing function of drones as an example, the drone simulation model presets the shape and flight control of the drone, and then adds the drone simulation model to the simulation scene. The added drone simulation model can fly smoothly at high, medium and high altitudes and low altitudes in the simulation scene.

[0061] The sensor simulation module is used to add the required sensors to the UAV simulation model and capture sensor data in the simulation scene.

[0062] The sensor model includes: camera, lidar, GPS, IMU, barometer, magnetometer, and distance sensor;

[0063] The sensor model supports personalized settings for the sensor, including: installation location, installation angle, and sensor coefficients, etc.

[0064] In particular, the sensor simulation module supports time and space synchronization of data acquisition from different sensors by setting a unified clock and a transformation matrix for the relative positions of different sensors. The transformation matrix for the relative positions of different sensors is directly input into the multi-sensor fusion algorithm for processing.

[0065] Taking the emergency landing function of a drone as an example, the emergency landing function needs to integrate data from both camera and lidar sensors. Therefore, the sensor simulation module adds camera and lidar sensor models to the drone simulation model in the simulation scene and sets the sensor coefficients.

[0066] The sensor simulation module, based on the camera installation position requirements of the emergency landing function, sets the camera's installation position and angle in the front (X)-right (Y)-down (Z) body coordinate system as follows: X = 0.50, Y = 0, Z = 0.10, Pitch = -90, Roll = 0, Yaw = 0. Furthermore, it also needs to set sensor coefficients such as image type, image resolution, and camera field of view to realize a downward-facing camera installed directly in front of the UAV, capturing visible light images in the simulated scene.

[0067] The sensor simulation module, based on the installation position requirements of the lidar for the emergency landing function, sets the installation position and angle of the lidar in the front (X)-right (Y)-down (Z) body coordinate system as follows: X=0, Y=0, Z=-0.50, pitch angle Pitch=-90, roll angle Roll=0, yaw angle Yaw=0. In addition, it also needs to set sensor coefficients such as the number of radar channels, detection range, number of point clouds generated per second, horizontal field of view, and vertical field of view to realize a lidar installed directly below the UAV to capture point cloud data in the vertical direction within the simulated scene.

[0068] The flight control module is used to generate the flight path of the UAV and to control the movement of the UAV. The control includes position control and speed control.

[0069] Based on the above technical solutions, it also includes: a communication module, a model inference module, a multi-sensor fusion algorithm module, and an intelligent function output module;

[0070] The communication module is used to transmit sensor data and flight control commands through the ROS publish and subscribe mechanism, wherein: the sensor data is transmitted from the sensor simulation module to the model inference module, and the control commands refer to flight control commands, which are transmitted from the intelligent function output module to the flight control module; it is also used to set the transmission frequency of sensor data to meet the real-time requirements of simulation.

[0071] The model inference module is used to iterate and generate a high-performance model based on sensor data, and then process the sensor data using the high-performance model. The high-performance model refers to the fact that different sensor data have different models, and different models have different evaluation indicators. The high-performance model is determined by comparing the evaluation indicators. The processing of sensor data refers to the fact that different sensor data can be processed in different ways, such as image data can be processed by image segmentation and target detection.

[0072] Taking the emergency landing function of a drone as an example, the model inference module iteratively generates an image segmentation model using a training image dataset. The image segmentation model receives the training image dataset and uses a large number of downward-looking camera images (the number of images is not fixed and can be empirical values ​​or implemented according to existing technology) for training, iteratively generating an image segmentation model with an accuracy of over 98% (accuracy is verified by a test image dataset). Using the image segmentation model with an accuracy of over 98%, the real-time transmitted urban simulation images are segmented to generate segmented images; among them, paved roads, grass, and other areas are landable areas, while houses and bushes are non-landable areas.

[0073] It should be noted that the model inference module is instantiated into different models for different intelligent function simulation tests. In the simulation test of the UAV emergency landing function, the model inference module needs to train the image data to obtain an image segmentation model.

[0074] The multi-sensor fusion algorithm module fuses data from different sensors based on the data processing results output by the model inference module.

[0075] The multi-sensor fusion algorithm module is part of the simulation test and verification platform. It takes the processing result of the model inference module as input and then fuses the data from different sensors.

[0076] Taking the emergency landing function as an example, the multi-sensor fusion algorithm module takes the processing result of the model inference module as input, receives the segmented image output by the model inference module, obtains the landable area and non-landable area in the segmented image, and then fuses the segmented image and point cloud data.

[0077] The intelligent function output module acquires the data fusion results output by the multi-sensor fusion algorithm module, outputs UAV control commands, and realizes intelligent functions.

[0078] Based on the data fusion results from the multi-sensor fusion algorithm module, the intelligent function output module sends UAV control commands to the flight control module to achieve the corresponding intelligent functions.

[0079] Taking the emergency landing function as an example, multi-sensor fusion is part of the simulation test and verification platform. Based on the fusion result obtained by fusing segmented images and point cloud data, the intelligent function output module identifies the landable area in the current environment, and then sends UAV control commands to the flight control module according to the identification result to realize the UAV emergency landing function.

[0080] like Figure 2 As shown, this invention presents a simulation testing and verification method for multi-sensor fusion in unmanned aerial vehicles (UAVs), which includes the following steps:

[0081] The aircraft parameters are pre-set in the simulation test and verification platform according to the simulation requirements;

[0082] Sensor parameters are pre-set in the simulation test and verification platform according to simulation requirements;

[0083] Flight route parameters are pre-set in the simulation test and verification platform according to simulation requirements;

[0084] The scene simulation module generates a three-dimensional virtual simulation scene, the aircraft simulation module adds a drone simulation model to the three-dimensional virtual simulation scene, and the sensor simulation module adds the required sensors to the drone simulation model.

[0085] The flight control module controls the movement of the UAV, which moves along a pre-set flight path in a three-dimensional virtual simulation scene.

[0086] Based on the above technical solution, the following steps are also included:

[0087] The sensor simulation module synchronously captures sensor data in a 3D virtual simulation scene;

[0088] The communication module sends sensor data to the model inference module;

[0089] The model inference module processes the input sensor data using a high-performance model.

[0090] The multi-sensor fusion algorithm module fuses the processed sensor data;

[0091] The intelligent function output module outputs drone control commands based on the results of multi-sensor data fusion, thereby realizing the intelligent functions of the drone.

[0092] Based on the above technical solution, the following steps are also included:

[0093] The communication module transmits the UAV control commands issued by the intelligent function output module to the flight control module, which then controls the aircraft to complete the simulation of the intelligent function.

[0094] Taking the emergency landing function of unmanned aerial vehicles (UAVs) as an example, its core function is to enhance the UAV's environmental perception capabilities through LiDAR and downward-looking cameras, and to integrate point cloud data and image data to actively identify landable areas in different unknown environments such as low altitude, mid-high altitude, and high altitude in urban scenarios. The simulation test verification process of the UAV emergency landing function in this invention is as follows:

[0095] (1) Pre-set the aircraft parameters in the simulation platform, including shape and flight control;

[0096] (2) Set the sensor parameters in the simulation platform in advance, including sensor type, sensor installation position, sensor installation angle, camera image capture type, captured image resolution, camera field of view, artifacts, number of lidar channels, detection range, and field of view;

[0097] (3) Set the drone flight path on the simulation platform;

[0098] (4) Run the scene simulation module and the aircraft simulation module to simulate a UAV with a downward-looking camera and a lidar in a three-dimensional virtual city scene;

[0099] (5) Run the flight control module to control the UAV to move in the three-dimensional virtual city scene according to the planned flight route;

[0100] (6) Run the sensor simulation module to simultaneously capture image data from the downward-looking camera and point cloud data from the lidar in a three-dimensional virtual city scene;

[0101] (7) Run the communication module, set the transmission frequency of image data and point cloud data, and publish them through ROS;

[0102] (8) Run the model inference module, subscribe to image data through ROS, input it into the image segmentation model to realize its iterative update, and use the iterated image segmentation module to segment the image data;

[0103] (9) Run the multi-sensor fusion algorithm module, subscribe to point cloud data through ROS, and fuse the segmented image with the point cloud data;

[0104] (10) Run the emergency landing function output module, identify the landable area in the current environment based on the fusion result of image and point cloud data, and generate UAV control commands according to the identification result;

[0105] (11) Run the communication module to send the UAV control commands issued by the emergency landing function output module to the flight control module, and control the UAV to achieve emergency landing autonomously.

[0106] The contents not described in detail in this specification are existing technologies known to those skilled in the art.

[0107] The above description is only a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. Any equivalent modifications or changes made by those skilled in the art based on the content disclosed in the present invention should be included in the scope of protection set forth in the claims.

Claims

1. A simulation test and verification platform for multi-sensor fusion in unmanned aerial vehicles (UAVs), characterized in that, include: Scene simulation module, UAV simulation module, sensor simulation module, flight control module, communication module, model inference module, multi-sensor fusion algorithm module, intelligent function output module; The scene simulation module uses Unreal Engine to generate uproject files for storing simulation scenes, providing three-dimensional virtual drone simulation scenes, and providing weather and scenario settings within the simulation scenes; The UAV simulation module is used to add UAV simulation models to the simulation scene, to provide UAV simulation models that conform to the kinematics and dynamics characteristics of the real aircraft, and to set the UAV shape and flight control through the UAV simulation model; The sensor simulation module is used to add the required sensors to the UAV simulation model and capture sensor data in the simulation scene. The sensor simulation module supports time and space synchronization of data acquisition from different sensors by setting a unified clock and a relative position transformation matrix for different sensors. The relative position transformation matrix for different sensors is directly input into the multi-sensor fusion algorithm module for processing. The flight control module is used to generate the flight path of the UAV and to control the movement of the UAV, including position control and speed control. The communication module is used to transmit sensor data and flight control commands through the ROS publish and subscribe mechanism, wherein: the sensor data is transmitted from the sensor simulation module to the model inference module, and the flight control commands are transmitted from the intelligent function output module to the flight control module; Used to set the transmission frequency of sensor data to meet the real-time requirements of simulation; The model inference module is used to iterate and generate a high-performance model based on sensor data, and then process the sensor data using the high-performance model. The high-performance model refers to the fact that different sensor data have different models, and different models have different evaluation indicators. The high-performance model is determined by comparing the evaluation indicators. The processing of sensor data refers to the use of different processing methods for different sensor data. The model inference module instantiates different models for simulation tests of different intelligent functions. The multi-sensor fusion algorithm module is part of the simulation test and verification platform, which fuses data from different sensors based on the data processing results output by the model inference module. The intelligent function output module acquires the data fusion results output by the multi-sensor fusion algorithm module, outputs UAV control commands, and realizes intelligent functions.

2. The simulation test and verification platform for multi-sensor fusion of unmanned aerial vehicles as described in claim 1, characterized in that, The simulation scene includes several typical terrain scenes, specifically divided into: basic terrain scene, basic feature scene, and basic natural scene. The basic terrain features include: villas, high-rise buildings, factories, commercial buildings, and power transmission towers; The basic feature scenes include: cars, pedestrians, streetlights, power lines, highways, paved roads, swimming pools, benches, and fences; The basic natural scenes include: mountains, forests, lakes, and grasslands; The meteorological and scenario settings include: rain, snow, dust storms, fog and haze, road surface humidity, snow accumulation on the road surface, and fallen leaves on the road surface; the scenarios refer to natural weather conditions, including: lighting conditions, rain and snow, dust storms and fog and haze, and natural wind.

3. The simulation test and verification platform for multi-sensor fusion of unmanned aerial vehicles as described in claim 1, characterized in that, The sensor models added to the UAV simulation model by the sensor simulation module include: camera, lidar, GPS, IMU, barometer, magnetometer, and distance sensor; The sensor model supports personalized settings for the sensor, including installation location, installation angle, and sensor coefficients.

4. A simulation testing and verification method for multi-sensor fusion in unmanned aerial vehicles (UAVs), characterized in that, The method using the simulation test and verification platform according to any one of claims 1-3 includes the following steps: Based on the simulation requirements, the drone parameters are pre-set in the simulation test and verification platform; Sensor parameters are pre-set in the simulation test and verification platform according to simulation requirements; Flight route parameters are pre-set in the simulation test and verification platform according to simulation requirements; The scene simulation module generates a 3D virtual simulation scene, the UAV simulation module adds a UAV simulation model to the 3D virtual simulation scene, and the sensor simulation module adds the required sensors to the UAV simulation model. The flight control module controls the movement of the UAV, which moves along a pre-set flight path in a three-dimensional virtual simulation scene.

5. The simulation testing and verification method for multi-sensor fusion of unmanned aerial vehicles as described in claim 4, characterized in that, It also includes the following steps: The sensor simulation module synchronously captures sensor data in a 3D virtual simulation scene; The communication module sends sensor data to the model inference module; The model inference module processes the input sensor data using a high-performance model. The multi-sensor fusion algorithm module fuses the processed sensor data; The intelligent function output module outputs drone control commands based on the results of multi-sensor data fusion, thereby realizing the intelligent functions of the drone.

6. The simulation testing and verification method for multi-sensor fusion of unmanned aerial vehicles as described in claim 5, characterized in that, It also includes the following steps: The communication module transmits the UAV control commands issued by the intelligent function output module to the flight control module, controlling the UAV to complete the intelligent function simulation.

Images