ROS-Gazebo-based airport ground support vehicle modeling and operation simulation method

Abstract

The invention discloses an airport ground support vehicle modeling and operation simulation method based on ROS-Gazebo. The method comprises the steps of obtaining an airport operation environment and an airport ground support vehicle model; building a physical model of an airport operation environment and an airport ground support vehicle; designing a motion model of the airport ground support vehicle; a vehicle simulation function package based on ROS-Gazebo is designed; constructing a two-dimensional grid map around the airport ground support vehicle; making an aircraft operation part data set and training an operation part identification network model based on YOLOv5; self-adaptive Monte Carlo positioning and a mobile navigation stack of the airport ground support vehicle are configured; and simulating the simulation process of different types of airport ground support vehicles in the airport environment. The performance of the automatic driving system in different situations can be comprehensively tested, and the adaptability and robustness of the system are improved. Meanwhile, the method has the advantages that the method is not limited by time and space conditions, training time can be shortened, training cost can be reduced, and safety and controllability are achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of autonomous driving simulation, and in particular relates to a modeling and operation simulation method for airport ground support vehicles based on ROS-Gazebo. Background Art

[0002] With the increase in the number of flights at large hub airports, the demand for improving the efficiency and safety of aircraft ground support operations is increasing. Compared with traditional manned airport ground support equipment, the advantages of unmanned ground support equipment are: it can reduce the labor intensity of personnel, reduce manual participation, improve the operating efficiency of ground support equipment, and also help airport companies reduce production costs and improve production efficiency. At the same time, unmanned airport ground support equipment uses a variety of sensors to perceive the operating environment in real time, and uses intelligent algorithms for dynamic planning and decision-making, which is conducive to increasing the reliability and safety of equipment driving and operation, and reducing aircraft damage caused by human factors and operational conflicts between equipment and aircraft.

[0003] In the process of developing and testing unmanned ground support vehicles at airports, there are many difficulties and risks in testing in actual scenarios, such as the complexity of road conditions, high testing costs, and uncontrollable safety. In order to avoid field experiments that may endanger the safety of airport operations, vehicle autonomous driving simulation technology provides a safe and controllable method. Through the simulation environment, researchers can conduct large-scale tests and experiments more economically and efficiently, flexibly simulate a variety of scenarios, reduce experimental costs and time investment, and comprehensively test the performance of the autonomous driving system in different scenarios, and improve the adaptability and robustness of the autonomous operation system of ground support vehicles. At the same time, creating a simulated and controllable virtual airport site, including traffic networks, taxiways, aprons, and possible moving obstacles (such as other operating vehicles and personnel, etc.), improves the realism of the virtual environment, and can also provide more accurate data and experimental basis for subsequent research. However, no relevant methods have been found so far. Summary of the invention

[0004] In order to solve the above problems, the purpose of the present invention is to provide an airport ground support vehicle modeling and operation simulation method based on ROS-Gazebo.

[0005] In order to achieve the above object, the airport ground support vehicle modeling and operation simulation method based on ROS-Gazebo provided by the present invention comprises the following steps performed in sequence:

[0006] Step 1) Use 3dMaxs and Blueender software to draw the airport operating environment, including common elements in the airport, such as ground markings, signboards, operating equipment, terminal buildings and aircraft. At the same time, for the target aircraft that the airport ground support vehicle is about to operate, add a high-recognition map to the operating part of the target aircraft; use SolidWorks software to accurately model the airport ground support vehicle in three dimensions. At the same time, for different airport ground support vehicles, including passenger elevators, baggage transport vehicles, and power supply vehicles, accurately draw the operating structures and required sensors installed on them to obtain the airport operating environment model and the airport ground support vehicle model;

[0007] Step 2) exporting the airport operating environment model obtained in step 1) into st l or dae format, and importing it into the custom model library of GAZEBO simulation software to obtain the environment model module; exporting the airport ground support vehicle model obtained in step 1) and generating a parameterized model file URFD in XML syntax format, and then modifying the URDF file to obtain the vehicle model module, and at the same time, for the sensor drawn in step 1), creating a plug-in to simulate the behavior of the sensor, and in the GAZEBO simulation software, integrating a variety of sensors on the vehicle model module, including IMU, GPS, laser radar and camera; then combining the above environment model module and vehicle model module in GAZEBO to complete the construction of the physical model of the airport operating environment and the airport ground support vehicle;

[0008] Step 3) Designing a motion model of the airport ground support vehicle; the motion model is divided into two parts, the first part is the vehicle chassis, and the vehicle chassis adopts the Ackerman vehicle chassis motion model; the second part is the upper operating structure;

[0009] Step 4) Design a vehicle simulation package based on ROS-Gazebo, which will integrate sensors and publish and subscribe to these sensor data through ROS topic nodes; in addition, the vehicle simulation package will contain control algorithms, including PID controllers and precise control logic for the job structure; the vehicle simulation package will also provide a user interface; finally, through detailed documentation and sample code, the vehicle simulation package will help users understand and use the simulation tool;

[0010] Step 5) Load the environment model module and vehicle model module obtained in step 2) into Gazebo, drive the airport ground support vehicle through the airport environment and collect necessary data in the simulation environment through the vehicle simulation function package designed in step 4), and use the gmap ing algorithm, combined with the odometer and lidar data, to build a two-dimensional grid map around the airport ground support vehicle in the Gazebo simulation environment;

[0011] Step 6) Create an aircraft operation part data set and train an operation part recognition network model based on YOLOv5;

[0012] Step 7) Configure the adaptive Monte Carlo positioning and move_base navigation stack for the airport ground support vehicle;

[0013] Step 8) expands on the basis of the vehicle simulation function package designed in step 4), designs a vehicle ground support operation simulation test based on the launch file for specific task requirements, integrates the environment model module and the vehicle model module, realizes visualization in the GAZEBO scene, loads the two-dimensional grid map in RVI Z and sets the target point, realizes autonomous navigation of the airport ground support vehicle according to the real-time data sent by the sensor and the control algorithm in the vehicle simulation function package, confirms the operation target through the operation part recognition network model constructed in step 6) when approaching the target aircraft, and controls the upper operation structure designed in step 3) to complete the ground support task, thereby simulating the simulation process of different types of airport ground support vehicles in the airport environment.

[0014] In step 3), the method of designing a motion model of an airport ground support vehicle; the motion model is divided into two parts, the first part is the vehicle chassis, and the vehicle chassis adopts the Ackerman vehicle chassis motion model; the second part is the operating structure of the upper body:

[0015] According to the Ackerman model theory, the motion equation of the airport ground support vehicle is constructed as shown in equations (1) and (2). The motion equation describes the speed relationship between the front and rear wheels of the airport ground support vehicle when turning.

[0016]

[0017] Where l is the distance between the front and rear axles of the airport ground support vehicle; θ is the front turning angle of the airport ground support vehicle; 2d is the distance between the two rear wheels of the airport ground support vehicle; v is the linear speed of the airport ground support vehicle; v L v is the linear speed of the left rear wheel of the airport ground support vehicle; R is the linear speed of the right rear wheel of the airport ground support vehicle; ω is the angular velocity of the airport ground support vehicle;

[0018] Apply the above motion equations to the vehicle chassis to ensure that the motion behavior of the airport ground support vehicle in the simulation environment matches the motion characteristics of the actual airport ground support vehicle, so as to achieve a unified transformation of the motion model of the airport ground support vehicle;

[0019] The second part is the operating structure of the superstructure. According to the operating characteristics of ground support vehicles at different airports, the motion model of the superstructure mechanical structure is defined to achieve accurate simulation of the operating process of ground support vehicles at different airports.

[0020] In step 6), the specific method of preparing the aircraft operation part data set and training the operation part recognition network model based on YOLOv5 is:

[0021] 6.1) Use cameras to capture images of different aircraft operating parts in different environments and conditions, ensuring that the collected images are diverse, including different lighting conditions, door angles, backgrounds and sizes;

[0022] 6.2) Using image annotation tools, mark the locations of aircraft operating parts on the above images by drawing rectangular bounding boxes and assigning category labels, including doors and refueling ports;

[0023] 6.3) Using image enhancement technology, the annotated images are processed by random rotation, scaling, cropping and color adjustment to obtain enhanced images. Each enhanced image corresponds to a .txt file. All enhanced images constitute the aircraft operation part dataset, and the aircraft operation part dataset is divided into training set, validation set and test set in a ratio of 6:2:2;

[0024] 6.4) Construct an original work part recognition network model based on YOLOv5, then input the training set into the original work part recognition network model for training. After the training is completed, use the validation set to evaluate the performance of the original work part recognition network model, and adjust the model parameters as needed; then input the test set for testing. If the original work part recognition network model performs well on the test set, save the weight file generated by the training, obtain the work part recognition network model and deploy it in practical applications;

[0025] 6.5) Create a node in the vehicle simulation function package obtained in step 4) to receive images from the camera on the airport ground support vehicle and transmit them to the above-mentioned work site recognition network model for processing.

[0026] The airport ground support vehicle modeling and operation simulation method based on ROS-Gazebo provided by the present invention has the following beneficial effects:

[0027] 1. On the one hand, the present invention takes into account the motion state of the vehicle's Ackerman chassis and the upper working structure, which can more accurately reflect the motion characteristics of the airport ground support vehicle chassis and the upper working device. On the other hand, the site model in the special environment of the airport is imported into the simulation environment to obtain a more realistic simulation of the working environment. Therefore, the two aspects improve the accuracy and authenticity of the simulation method.

[0028] 2. Using ROS as the underlying method and utilizing the distributed software architecture, a complete set of airport ground support vehicle autonomous driving simulation methods is built, integrating the loosely coupled and highly cohesive software functional modules into a unified method, thus improving the versatility, realism, and scalability of the system.

[0029] 3. The airport ground support vehicle automatic docking simulation test platform established by the present invention can provide a safe and controllable method, avoiding the risk of endangering the safety of airport operations by conducting field experiments at actual airport sites. Using the simulation environment, researchers can conduct large-scale tests and experiments more economically and efficiently, and can simulate these various scenarios more flexibly, reducing the cost and time investment of the experiment. It can also comprehensively test the performance of the autonomous driving system in different scenarios and improve the adaptability and robustness of the system. At the same time, this method has the advantages of not being restricted by time and space conditions, shortening training time, reducing training costs, and being safe and controllable. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a flow chart of the airport ground support vehicle modeling and operation simulation method based on ROS-Gazebo provided by the present invention. DETAILED DESCRIPTION

[0031] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0032] like Figure 1 As shown, the airport ground support vehicle modeling and operation simulation method based on ROS-Gazebo provided by the present invention includes the following steps performed in sequence:

[0033] Step 1) Use 3dMaxs and Blender software to draw the airport operating environment, including common elements in the airport, such as ground markings, signboards, operating equipment, terminal buildings and aircraft. At the same time, for the target aircraft that the airport ground support vehicle is about to operate, add high-recognition maps to the operating parts of the target aircraft to enhance the realism and visual effects of the simulation environment, and provide clear targets for subsequent camera visual recognition simulation; Use SolidWorks software to accurately model the airport ground support vehicle in three dimensions to ensure that the geometric accuracy and physical properties of the model meet the specifications of the actual airport ground support vehicle. At the same time, for different airport ground support vehicles, including passenger elevators, baggage transport vehicles, power supply vehicles, etc., accurately draw the operating structures and required sensors installed on them to obtain the airport operating environment model and airport ground support vehicle model;

[0034] Step 2) exporting the airport operating environment model obtained in step 1) into stl or dae format, and importing it into the custom model library of the GAZEBO simulation software to obtain the environment model module; exporting the airport ground support vehicle model obtained in step 1) and generating a parameterized model file URFD in XML syntax format, then modifying the URDF file to obtain the vehicle model module, and at the same time, for the sensor drawn in step 1), creating a plug-in to simulate the behavior of the sensor, in the GAZEBO simulation software, integrating a variety of sensors on the vehicle model module, including IMU (inertial measurement unit), GPS, laser radar (L IDAR) and camera, etc., for real-time acquisition of the speed, posture, position and environmental information of the airport ground support vehicle, so that the sensor data can be received in the subsequent simulation process; then combining the above-mentioned environment model module and vehicle model module in GAZEBO to complete the construction of the physical model of the airport operating environment and the airport ground support vehicle;

[0035] Step 3) Designing a motion model of the airport ground support vehicle; the motion model is divided into two parts, the first part is the vehicle chassis, and the vehicle chassis adopts the Ackerman vehicle chassis motion model;

[0036] According to the Ackerman model theory, the motion equation of the airport ground support vehicle is constructed as shown in equations (1) and (2). The motion equation describes the speed relationship between the front and rear wheels of the airport ground support vehicle when turning.

[0037]

[0038] Where l is the distance between the front and rear axles of the airport ground support vehicle; θ is the front turning angle of the airport ground support vehicle; 2d is the distance between the two rear wheels of the airport ground support vehicle; v is the linear speed of the airport ground support vehicle; v L v is the linear speed of the left rear wheel of the airport ground support vehicle; R is the linear speed of the right rear wheel of the airport ground support vehicle; ω is the angular velocity of the airport ground support vehicle;

[0039] Apply the above motion equations to the vehicle chassis to ensure that the motion behavior of the airport ground support vehicle in the simulation environment matches the motion characteristics of the actual airport ground support vehicle, so as to achieve a unified transformation of the motion model of the airport ground support vehicle;

[0040] The second part is the operating structure of the upper body. According to the operating characteristics of ground support vehicles at different airports, the motion model of the upper body mechanical structure is defined to achieve accurate simulation of the operating process of ground support vehicles at different airports.

[0041] Step 4) Design a vehicle simulation function package based on ROS-Gazebo to achieve accurate acquisition of the speed, posture and position of the airport ground support vehicle in the Gazebo simulation environment, while providing motion control of the vehicle chassis and operating structure. It will integrate sensors and publish and subscribe to these sensor data through ROS topic nodes; in addition, the vehicle simulation function package will contain control algorithms, such as PID controllers, and precise control logic for the operating structure; in order to ensure that the simulation is synchronized with reality, it will implement anomaly detection and processing mechanisms, and adopt a modular design for easy maintenance and upgrades. The vehicle simulation function package will also provide a user interface to facilitate users to view the status of airport ground support vehicles in real time, send control commands, and adjust simulation parameters. Finally, through detailed documentation and sample code, the vehicle simulation function package will help users understand and use simulation tools, so as to test and verify the control strategies and behaviors of airport ground support vehicles without the need for expensive and dangerous real-world experiments;

[0042] Step 5) Load the environment model module and vehicle model module obtained in step 2) into Gazebo, drive the airport ground support vehicle through the airport environment and collect necessary data in the simulation environment through the vehicle simulation function package designed in step 4), and use the gmapping algorithm to combine the odometer and laser radar (L IDAR) data to build a two-dimensional grid map around the airport ground support vehicle in the Gazebo simulation environment. The gmapping algorithm estimates the position of the airport ground support vehicle and builds an environmental map at the same time by fusing sensor data to establish a detailed two-dimensional grid map. The map not only includes the geometry of the environment, but also includes obstacles, road markings and other important features. The obtained two-dimensional grid map is used for path planning and autonomous driving function testing. These tests will verify the performance of the autonomous driving system and provide feedback for further optimization and development.

[0043] Step 6) Create an aircraft operation part data set and train an operation part recognition network model based on YOLOv5;

[0044] The specific method is as follows:

[0045] 6.1) Use cameras to capture images of different aircraft operating parts in different environments and conditions, ensuring that the collected images are diverse, including different lighting conditions, door angles, backgrounds and sizes;

[0046] 6.2) Use image annotation tools to mark the locations of aircraft operating parts on the above images, including doors, refueling ports, etc., by drawing rectangular bounding boxes and specifying category labels;

[0047] 6.3) Using image enhancement technology, the annotated images are processed by random rotation, scaling, cropping and color adjustment to obtain enhanced images, so as to increase the diversity of images and the generalization ability of the model, and ensure that all annotations meet the requirements of YOLOv5. Each enhanced image corresponds to a .txt file. All enhanced images constitute the aircraft operation part data set, and the aircraft operation part data set is divided into training set, verification set and test set in a ratio of 6:2:2;

[0048] 6.4) Construct an original work part recognition network model based on YOLOv5, and then input the training set into the original work part recognition network model for training. After the training is completed, use the validation set to evaluate the performance of the original work part recognition network model, and adjust the model parameters as needed; then input the test set for testing. If the original work part recognition network model performs well on the test set, save the weight file generated by the training, obtain the work part recognition network model and deploy it in practical applications.

[0049] 6.5) Create a node in the vehicle simulation function package obtained in step 4) to receive images from the camera on the airport ground support vehicle and transmit them to the above-mentioned operation position recognition network model for processing; by deploying the yo l ov5_ros node and using the images captured by the camera on the airport ground support vehicle, the system can recognize and locate the operation position in real time;

[0050] Step 7) Configure the vehicle's adaptive Monte Carlo positioning (amc l) and move_base navigation stack. AMC l is used to estimate the position and posture information of the airport ground support vehicle in the map coordinate system, and provide TF transformation from the airport ground support vehicle to the odometer and map to ensure the accuracy of the airport ground support vehicle's positioning; at the same time, the move_base navigation stack controls the movement of the vehicle chassis according to the preset target points in the two-dimensional grid map generated in step 5), and implements global and local path planning by configuring the cost map and local planner to ensure that the airport ground support vehicle can reach the destination safely and efficiently; during the entire movement process, the system will continuously feedback the airport ground support vehicle's posture and target point status information for real-time monitoring and adjustment. Through the integration and testing of these technologies, the airport ground support vehicle can achieve reliable autonomous navigation functions in complex airport ground operation environments, improving the level of automation and safety of operations;

[0051] Step 8) expands on the basis of the vehicle simulation function package designed in step 4), designs a vehicle ground support operation simulation test based on the launch file for specific task requirements, integrates the environment model module and the vehicle model module, realizes visualization in the GAZEBO scene, loads the two-dimensional grid map in RVI Z and sets the target point, realizes autonomous navigation of the airport ground support vehicle according to the real-time data sent by the sensor and the control algorithm in the vehicle simulation function package, confirms the operation target through the operation part recognition network model constructed in step 6) when approaching the target aircraft, and controls the upper operation structure designed in step 3) to complete the ground support task, thereby simulating the simulation process of different types of airport ground support vehicles in the airport environment.

Claims

1. A modeling and operation simulation method for airport ground support vehicles based on ROS-Gazebo, characterized by: The ROS-Gazebo-based airport ground support vehicle modeling and operation simulation method includes the following steps performed in sequence: Step 1) Use 3dMaxs and Blender software to draw the airport operating environment, including common elements in the airport, such as ground markings, signboards, operating equipment, terminal buildings and aircraft. At the same time, for the target aircraft that the airport ground support vehicle is about to operate, add highly recognizable maps to the operating parts of the target aircraft; use SolidWorks software to accurately model the airport ground support vehicle in three dimensions. At the same time, for different airport ground support vehicles, including passenger elevators, baggage transport vehicles, and power supply vehicles, accurately draw the operating structures and required sensors installed on them to obtain the airport operating environment model and the airport ground support vehicle model; Step 2) exporting the airport operating environment model obtained in step 1) into stl or dae format, and importing it into the custom model library of the GAZEBO simulation software to obtain the environment model module; exporting the airport ground support vehicle model obtained in step 1) and generating a parameterized model file URFD in xml syntax format, and then modifying the URDF file to obtain the vehicle model module, and at the same time, for the sensor drawn in step 1), creating a plug-in to simulate the behavior of the sensor, and in the GAZEBO simulation software, integrating a variety of sensors on the vehicle model module, including IMU, GPS, laser radar and camera; then combining the above environment model module and vehicle model module in GAZEBO to complete the construction of the physical model of the airport operating environment and the airport ground support vehicle; Step 3) Designing a motion model of the airport ground support vehicle; the motion model is divided into two parts, the first part is the vehicle chassis, and the vehicle chassis adopts the Ackerman vehicle chassis motion model; the second part is the upper operating structure; Step 4) Design a vehicle simulation package based on ROS-Gazebo, which will integrate sensors and publish and subscribe to these sensor data through ROS topic nodes; in addition, the vehicle simulation package will contain control algorithms, including PID controllers and precise control logic for the job structure; the vehicle simulation package will also provide a user interface; finally, through detailed documentation and sample code, the vehicle simulation package will help users understand and use the simulation tool; Step 5) Load the environment model module and vehicle model module obtained in step 2) into Gazebo, drive the airport ground support vehicle through the airport environment and collect necessary data in the simulation environment through the vehicle simulation function package designed in step 4), and use the gmapping algorithm, combined with the odometer and lidar data, to build a two-dimensional grid map around the airport ground support vehicle in the Gazebo simulation environment; Step 6) Create an aircraft operation part data set and train an operation part recognition network model based on YOLOv5; Step 7) Configure the adaptive Monte Carlo positioning and move_base navigation stack for the airport ground support vehicle; Step 8) expands on the basis of the vehicle simulation function package designed in step 4), designs a vehicle ground support operation simulation test based on the launch file for specific task requirements, integrates the environment model module and the vehicle model module, realizes visualization in the GAZEBO scene, loads the two-dimensional grid map in RVIZ and sets the target point, realizes autonomous navigation of the airport ground support vehicle according to the real-time data sent by the sensor and the control algorithm in the vehicle simulation function package, confirms the operation target when approaching the target aircraft through the operation part identification network model constructed in step 6), and controls the upper operation structure designed in step 3) to complete the ground support task, thereby simulating the simulation process of different types of airport ground support vehicles in the airport environment.

2. The airport ground support vehicle modeling and operation simulation method based on ROS-Gazebo according to claim 1 is characterized by: In step 3), the method of designing a motion model of an airport ground support vehicle; the motion model is divided into two parts, the first part is the vehicle chassis, and the vehicle chassis adopts the Ackerman vehicle chassis motion model; the second part is the operating structure of the upper body: According to the Ackerman model theory, the motion equation of the airport ground support vehicle is constructed as shown in equations (1) and (2). The motion equation describes the speed relationship between the front and rear wheels of the airport ground support vehicle when turning. Where l is the distance between the front and rear axles of the airport ground support vehicle; θ is the front turning angle of the airport ground support vehicle; 2d is the distance between the two rear wheels of the airport ground support vehicle; v is the linear speed of the airport ground support vehicle; v L v is the linear speed of the left rear wheel of the airport ground support vehicle; R is the linear speed of the right rear wheel of the airport ground support vehicle; ω is the angular velocity of the airport ground support vehicle; Apply the above motion equations to the vehicle chassis to ensure that the motion behavior of the airport ground support vehicle in the simulation environment matches the motion characteristics of the actual airport ground support vehicle, so as to achieve a unified transformation of the motion model of the airport ground support vehicle; The second part is the operating structure of the superstructure. According to the operating characteristics of ground support vehicles at different airports, the motion model of the superstructure mechanical structure is defined to achieve accurate simulation of the operating process of ground support vehicles at different airports.

3. The airport ground support vehicle modeling and operation simulation method based on ROS-Gazebo according to claim 1 is characterized in that: In step 6), the specific method of preparing the aircraft operation part data set and training the operation part recognition network model based on YOLOv5 is: 6.1) Use cameras to capture images of different aircraft operating parts in different environments and conditions, ensuring that the collected images are diverse, including different lighting conditions, door angles, backgrounds and sizes; 6.2) Using image annotation tools, mark the locations of aircraft operating parts on the above images by drawing rectangular bounding boxes and assigning category labels, including doors and refueling ports; 6.3) Using image enhancement technology, the annotated images are processed by random rotation, scaling, cropping and color adjustment to obtain enhanced images. Each enhanced image corresponds to a .txt file. All enhanced images constitute the aircraft operation part dataset, and the aircraft operation part dataset is divided into training set, validation set and test set in a ratio of 6:2:2; 6.4) Construct an original work part recognition network model based on YOLOv5, then input the training set into the original work part recognition network model for training. After the training is completed, use the validation set to evaluate the performance of the original work part recognition network model, and adjust the model parameters as needed; then input the test set for testing. If the original work part recognition network model performs well on the test set, save the weight file generated by the training, obtain the work part recognition network model and deploy it in practical applications; 6.5) Create a node in the vehicle simulation function package obtained in step 4) to receive images from the camera on the airport ground support vehicle and transmit them to the above-mentioned work site recognition network model for processing.

Images