# Gazebo 多机仿真

本文介绍如何使用 Gazebo 仿真器配合软件在环仿真进行多机仿真(仅适用于 Linux )。 A different approach is used for simulation with and without ROS.

# Multiple Vehicle with Gazebo (No ROS)

To simulate multiple iris or plane vehicles in Gazebo use the following commands in the terminal (from the root of the Firmware tree):

Tools/gazebo_sitl_multiple_run.sh [-m <model>] [-n <number_of_vehicles>] [-w <world>] [-s <script>] [-t <target>] [-l <label>]
  • <model>: The vehicle type/model to spawn, e.g.: iris (default), plane, standard_vtol, rover, r1_rover typhoon_h480.

  • <number_of_vehicles>: The number of vehicles to spawn. Default is 3. Maximum is 255.

  • <world>: The world that the vehicle should be spawned into, e.g.: empty (default)

  • <script>: Spawn multiple vehicles of different types (overriding the values in -m and -n). 例如:

    -s "iris:3,plane:2,standard_vtol:3"
    • Supported vehicle types are: iris, plane, standard_vtol, rover, r1_rover typhoon_h480.
    • The number after the colon indicates the number of vehicles (of that type) to spawn.
    • Maximum number of vehicles is 255.
  • <target>: build target, e.g: px4_sitl_default (default), px4_sitl_rtps

  • <label> : specific label for model, e.g: rtps

Each vehicle instance is allocated a unique MAVLink system id (1, 2, 3, etc.). Vehicle instances are accessed from sequentially allocated PX4 remote UDP ports: 14540 - 14548 (additional instances are all accessed using the same remote UDP port: 14549).

{% youtube %}

# Video: Multiple Multicopter (Iris)


# Video: Multiple Plane

@{% endyoutube %}

# Video: Multiple VTOL

@{% youtube %}

# Build and Test (RTPS/DDS)

To simulate multiple vehicles based on RTPS/DDS in Gazebo, use the gazebo_sitl_multiple_run.sh command in the terminal with the -t px4_sitl_rtps option from the root of the PX4-Autopilot tree (as described above). Here we will use the -t px4_sitl_rtps option, which sets that we will use RTPS for communicating with PX4 rather than the MAVLink Simulation API. This builds and runs the iris model and by default also starts the microRTPS client (you can change the model using the -m parameter).


You will need to have installed or eProsima Fast DDS or ROS 2 Foxy or above and the micrortps_agent should be run in the different terminals for each vehicle. For more information see: RTPS/DDS Interface: PX4-Fast RTPS(DDS) Bridge, for how to use the interaction with non-ROS2 DDS participant applications, or ROS 2 User Guide (PX4-ROS 2 Bridge), for interfacing with ROS2 nodes.

{% youtube %}

  1. 克隆 PX4 固件源码, 然后编译 SITL 代码:

    cd Firmware_clone
    git submodule update --init --recursive
    DONT_RUN=1 make px4_sitl_rtps gazebo
  2. Build the micrortps_agent

  3. Run gazebo_sitl_multiple_run.sh. For example, to spawn 4 vehicles, run:

    ./Tools/gazebo_sitl_multiple_run.sh -t px4_sitl_rtps -m iris -n 4


Each vehicle instance is allocated a unique MAVLink system id (1, 2, 3, etc.), can receive data from a unique remote UDP port (2019, 2021, 2023, etc.), and transmit data to UDP port (2020, 2022, 2024, etc.).


  1. Run micrortps_agent. For example, to connect 4 vehicles, run:

    micrortps_agent -t UDP -r 2020 -s 2019 -n vhcl0 &
    micrortps_agent -t UDP -r 2022 -s 2021 -n vhcl1 &
    micrortps_agent -t UDP -r 2024 -s 2023 -n vhcl2 &
    micrortps_agent -t UDP -r 2026 -s 2025 -n vhcl3 &


In order to communicate with a specific instance of PX4 using ROS2, you must use the -n <namespace> option. For example, running micrortps_agent -t UDP -r 2020 -s 2019 -n vhcl0 will result in the agent publishing all its topics with the namespace prefix /vhcl0 (eg. sensor_combined data from vhcl0 will be published on the topic /vhcl0/fmu/sensor_combined/out, while if one wants to send commands to the same vehicle, it has to publish to topic /vhcl0/fmu/vehicle_command/in). You can then subscribe and publish to just that vehicle's topics.

# Multiple Vehicles with ROS and Gazebo

This example demonstrates a setup that opens the Gazebo client GUI showing two Iris vehicles in an empty world. You can then control the vehicles with QGroundControl and MAVROS in a similar way to how you would manage a single vehicle.

# 仿真前准备


At time of writing this is Ubuntu 18.04 with ROS Melodic/Gazebo 9. See also Gazebo Simulation.

# 开始仿真

{% endyoutube %}

  1. Clone the PX4/PX4-Autopilot code, then build the SITL code

    cd Firmware_clone
    git submodule update --init --recursive
    DONT_RUN=1 make px4_sitl_default gazebo
  2. Source your environment:

    source Tools/setup_gazebo.bash $(pwd) $(pwd)/build/px4_sitl_default
    export ROS_PACKAGE_PATH=$ROS_PACKAGE_PATH:$(pwd):$(pwd)/Tools/sitl_gazebo
  3. 运行启动文件:

    roslaunch px4 multi_uav_mavros_sitl.launch


You can specify gui:=false in the above roslaunch to launch Gazebo without its UI.

The tutorial example opens the Gazebo client GUI showing two Iris vehicles in an empty world.

You can control the vehicles with QGroundControl or MAVROS in a similar way to how you would manage a single vehicle:

  • QGroundControl 中有一个下拉选项,你可以选择指定的飞行器。
  • MAVROS 要求你在 topic/service 路径之前包含合适的命名空间,(例如,你会用到 /uav1/mavros/mission/push )。

# 发生了什么?

For each simulated vehicle, the following is required:

  • Gazebo model: This is defined as xacro file in PX4-Autopilot/Tools/sitl_gazebo/models/rotors_description/urdf/<model>_base.xacro see here (opens new window). Currently, the model xacro file is assumed to end with base.xacro. 此模型应该有一个名为 mavlink_udp_port 的参数, 该参数定义了与 px4 节点通信的 udp 端口。 模型的 xacro 文件将用于生成包含您选择的 udp 端口的 urdf 模型。 若要定义 udp 端口,请在每个飞行器的启动文件中设置 mavlink_udp_port,请参阅例子here (opens new window)


If you are using the same vehicle model, you don't need a separate xacro file for each vehicle. The same xacro file is adequate.

  • PX4 node: This is the SITL PX4 app. It communicates with the simulator, Gazebo, through the same UDP port defined in the Gazebo vehicle model, i.e. mavlink_udp_port. 要在 px4 sitl 应用程序端设置 udp 端口, 您需要在启动文件中设置 SITL_UDP_PRT 参数, 以匹配前面讨论的 mavlink_udp_port, 请参阅 here (opens new window)。 启动文件中的开始文件路径由参数 vehicleID产生,参考这里 (opens new window)。 The MAV_SYS_ID for each vehicle in the startup file, see here (opens new window), should match the ID for that vehicle in the launch file here (opens new window). 这样能够帮助你确保启动文件和开始文件中的设置相同。

  • MAVROS node(可选): 如果要通过 ros 控制车辆, 可以在启动文件中运行一个单独的 mavros 节点, 请参阅 here (opens new window), 以便连接到 px4 sitl 应用程序。 您需要在启动文件中一些特殊的端口上启动 mavlink 流, 请参阅 这里 (opens new window)。 这些特殊端口需要与launch文件中为MAVROS节点设置的相符合。参考这里 (opens new window)

构建一个示例设置, 请按照以下步骤操作:

  • 在gazebo中加载一个世界

      <!-- Gazebo sim -->
      <include file="$(find gazebo_ros)/launch/empty_world.launch">
          <arg name="gui" value="$(arg gui)"/>
          <arg name="world_name" value="$(arg world)"/>
          <arg name="debug" value="$(arg debug)"/>
          <arg name="verbose" value="$(arg verbose)"/>
          <arg name="paused" value="$(arg paused)"/>
  • 对于每个飞行器来说

    • 从 xacro 创建 urdf 模型, 加载gazebo模型并运行 px4 sitl 应用程序实例

        <!-- PX4 SITL and vehicle spawn -->
        <include file="$(find px4)/launch/single_vehicle_spawn.launch">
            <arg name="x" value="0"/>
            <arg name="y" value="0"/>
            <arg name="z" value="0"/>
            <arg name="R" value="0"/>
            <arg name="P" value="0"/>
            <arg name="Y" value="0"/>
            <arg name="vehicle" value="$(arg vehicle)"/>
            <arg name="rcS" value="$(find px4)/posix-configs/SITL/init/$(arg est)/$(arg vehicle)_$(arg ID)"/>
            <arg name="mavlink_tcp_port" value="4560"/>
            <arg name="ID" value="$(arg ID)"/>
    • 运行mavros节点

        <!-- MAVROS -->
        <include file="$(find mavros)/launch/px4.launch">
            <arg name="fcu_url" value="$(arg fcu_url)"/>
            <arg name="gcs_url" value=""/>
            <arg name="tgt_system" value="$(arg ID)"/>
            <arg name="tgt_component" value="1"/>


The complete block for each vehicle is enclosed in a set of <group> tags to separate the ROS namespaces of the vehicles.

To add a third iris to this simulation there are two main components to consider:

  • UAV3 添加到multi_uav_mavros_sitl.launch
    • 复制已经存在的四旋翼(UAV1 或者 UAV2)
    • ID 改为 3
    • 与gazebo的通信,选择一个不同的 mavlink_udp_port端口
    • MAVROS通信端口选择是通过在fcu_url 中修改两个端口号。
  • 创建一个开始文件,并按照如下方式修改:
    • 复制已存在的iris rcs启动文件,(iris_1iris_2) ,重命名为iris_3
    • MAV_SYS_ID 值改为3
    • SITL_UDP_PRT 的值与 mavlink_udp_port相一致。
    • 第一个mavlink start 端口和mavlink stream端口值设置为相同值,用于和QGC通信。
    • 第二个mavlink start 端口值应与启动文件 fcu_url 中的值一致。


Be aware of which port is src and dst for the different endpoints.

# Multiple Vehicles using SDF Models

然后, 你可以使用 QGroundControl 和 MAVROS 控制多机,控制方式与单机类似。


  1. Install xmlstarlet from your Linux terminal:
    sudo apt install xmlstarlet
  2. Use roslaunch with the multi_uav_mavros_sitl_sdf.launch launch file:
    roslaunch multi_uav_mavros_sitl_sdf.launch vehicle:=<model_file_name>


Note that the vehicle model file name argument is optional (vehicle:=<model_file_name>); if omitted the plane model (opens new window) will be used by default.

启动文件 multi_uav_mavros_sitl.launch做了以下内容,

要在此模拟中添加第三个iris四旋翼, 需要考虑两个主要部分:

  1. You can choose to do either of:
    • modify the single_vehicle_spawn_sdf.launch file to point to the location of your model by changing the line below to point to your model:
      $(find px4)/Tools/sitl_gazebo/models/$(arg vehicle)/$(arg vehicle).sdf


Ensure you set the vehicle argument even if you hardcode the path to your model.

  • copy your model into the folder indicated above (following the same path convention).
  1. The vehicle argument is used to set the PX4_SIM_MODEL environment variable, which is used by the default rcS (startup script) to find the corresponding startup settings file for the model. Within PX4 these startup files can be found in the PX4-Autopilot/ROMFS/px4fmu_common/init.d-posix/ directory. For example, here is the plane model's startup script (opens new window). For this to work, the PX4 node in the launch file is passed arguments that specify the rcS file (etc/init.d/rcS) and the location of the rootfs etc directory ($(find px4)/build_px4_sitl_default/etc). For simplicity, it is suggested that the startup file for the model be placed alongside PX4's in PX4-Autopilot/ROMFS/px4fmu_common/init.d-posix/.

# 其他资源