[ROS study notes 13] TF coordinate transformation in ROS

[ROS study notes 13] TF coordinate transformation in ROS

Written in the front, this series of notes refers to the tutorial of AutoLabor, the specific project address is here

foreword

There are some practical tools built in ROS. Through these tools, a certain function or debugging program can be realized conveniently and quickly, so as to improve the development efficiency. This chapter mainly introduces the following components built in ROS:

• TF coordinate transformation, realize the conversion between different types of coordinate systems;
• rosbag is used to record the execution process of ROS nodes and can replay the process;
• The rqt toolbox integrates a variety of graphical debugging tools.

Learning objectives expected to be achieved in this chapter:

• Understand the concept and application scenarios of TF coordinate transformation;
• Able to complete the TF case independently: the little turtle follows;
• Recording and playback can be realized by using the rosbag command or encoding;
• Be able to skillfully use the graphical tools in rqt.

Case demonstration: the little turtle follows the implementation, this case is a built-in case in ros , enter the start command in the terminal

roslaunch turtle_tf2 turtle_tf2_demo_cpp.launch

roslaunch turtle_tf2 turtle_tf2_demo.launch

The keyboard can control the movement of one turtle, and the other follows the movement.

Example result:

---

1. Static coordinate transformation

The so-called static coordinate transformation means that the relative position between two coordinate systems is fixed.

Description of Requirement:

There is an existing robot model. The core structure includes the main body and the radar, each corresponding to a coordinate system. The origin of the coordinate system is located at the physical center of the main body and the radar respectively. The known displacement relationship between the origin of the radar and the origin of the main body is as follows: x 0.2 y0.0 z0. 5. The current radar detects an obstacle. The coordinates of the obstacle in the radar coordinate system are (2.0 3.0 5.0). What is the coordinate of the obstacle relative to the subject?

Result demo:

Realize analysis:

1. The relative relationship of the coordinate system can be published by the publisher
2. Subscriber, subscribe to the relative relationship of the published coordinate system, and then pass in the coordinate point information (can be hard-coded), and then realize the coordinate transformation with the help of tf, and output the result

**Implementation process:**C++ is consistent with Python implementation process

1. Create a new feature package and add dependencies
2. Write the publisher implementation
3. Write the Subscriber Implementation
4. Execute and view the results

---

Solution A: C++ implementation

1. Create a feature package

Create a project feature package that depends ontf2、tf2_ros、tf2_geometry_msgs、roscpp rospy std_msgs geometry_msgs

2. Publisher

/* 
    静态坐标变换发布方:
        发布关于 laser 坐标系的位置信息 

    实现流程:
        1.包含头文件
        2.初始化 ROS 节点
        3.创建静态坐标转换广播器
        4.创建坐标系信息
        5.广播器发布坐标系信息
        6.spin()
*/


// 1.包含头文件
#include "ros/ros.h"
#include "tf2_ros/static_transform_broadcaster.h"
#include "geometry_msgs/TransformStamped.h"
#include "tf2/LinearMath/Quaternion.h"

int main(int argc, char *argv[])
{
    
    
    setlocale(LC_ALL,"");
    // 2.初始化 ROS 节点
    ros::init(argc,argv,"static_brocast");
    // 3.创建静态坐标转换广播器
    tf2_ros::StaticTransformBroadcaster broadcaster;
    // 4.创建坐标系信息
    geometry_msgs::TransformStamped ts;
    //----设置头信息
    ts.header.seq = 100;
    ts.header.stamp = ros::Time::now();
    ts.header.frame_id = "base_link";
    //----设置子级坐标系
    ts.child_frame_id = "laser";
    //----设置子级相对于父级的偏移量
    ts.transform.translation.x = 0.2;
    ts.transform.translation.y = 0.0;
    ts.transform.translation.z = 0.5;
    //----设置四元数:将 欧拉角数据转换成四元数
    tf2::Quaternion qtn;
    qtn.setRPY(0,0,0);
    ts.transform.rotation.x = qtn.getX();
    ts.transform.rotation.y = qtn.getY();
    ts.transform.rotation.z = qtn.getZ();
    ts.transform.rotation.w = qtn.getW();
    // 5.广播器发布坐标系信息
    broadcaster.sendTransform(ts);
    ros::spin();
    return 0;
}

The configuration file is omitted here.

Example effect:

use

rviz

Start the graphical interface, and then you can view the results:

3. Subscriber

/*  
    订阅坐标系信息，生成一个相对于 子级坐标系的坐标点数据，转换成父级坐标系中的坐标点

    实现流程:
        1.包含头文件
        2.初始化 ROS 节点
        3.创建 TF 订阅节点
        4.生成一个坐标点(相对于子级坐标系)
        5.转换坐标点(相对于父级坐标系)
        6.spin()
*/
//1.包含头文件
#include "ros/ros.h"
#include "tf2_ros/transform_listener.h"
#include "tf2_ros/buffer.h"
#include "geometry_msgs/PointStamped.h"
#include "tf2_geometry_msgs/tf2_geometry_msgs.h" //注意: 调用 transform 必须包含该头文件

int main(int argc, char *argv[])
{
    
    
    setlocale(LC_ALL,"");
    // 2.初始化 ROS 节点
    ros::init(argc,argv,"tf_sub");
    ros::NodeHandle nh;
    // 3.创建 TF 订阅节点
    tf2_ros::Buffer buffer;
    tf2_ros::TransformListener listener(buffer);

    ros::Rate r(1);
    while (ros::ok())
    {
    
    
    // 4.生成一个坐标点(相对于子级坐标系)
        geometry_msgs::PointStamped point_laser;
        point_laser.header.frame_id = "laser";
        point_laser.header.stamp = ros::Time::now();
        point_laser.point.x = 2.0;
        point_laser.point.y = 3.0;
        point_laser.point.y = 5.0;
    // 5.转换坐标点(相对于父级坐标系)
        //新建一个坐标点，用于接收转换结果  
        //--------------使用 try 语句或休眠，否则可能由于缓存接收延迟而导致坐标转换失败------------------------
        try
        {
    
    
            geometry_msgs::PointStamped point_base;
            point_base = buffer.transform(point_laser,"base_link");
            ROS_INFO("转换后的数据:(%.2f,%.2f,%.2f),参考的坐标系是:%s",point_base.point.x,point_base.point.y,point_base.point.z,point_base.header.frame_id.c_str());

        }
        catch(const std::exception& e)
        {
    
    
            // std::cerr << e.what() << '\n';
            ROS_INFO("程序异常.....");
        }


        r.sleep();  
        ros::spinOnce();
    }


    return 0;
}

The configuration file is omitted here.

4. Execution

You can use the command line or the launch file to start the publishing node and the subscribing node respectively. If there is no abnormality in the program, the console will output the coordinate conversion result.

The result of execution:

---

Option B: Python implementation

1. Create a feature package

Create a project feature package that depends ontf2、tf2_ros、tf2_geometry_msgs、roscpp rospy std_msgs geometry_msgs

2. Publisher

#! /usr/bin/env python
"""  
    静态坐标变换发布方:
        发布关于 laser 坐标系的位置信息 
    实现流程:
        1.导包
        2.初始化 ROS 节点
        3.创建 静态坐标广播器
        4.创建并组织被广播的消息
        5.广播器发送消息
        6.spin
"""
# 1.导包
import rospy
import tf2_ros
import tf
from geometry_msgs.msg import TransformStamped

if __name__ == "__main__":
    # 2.初始化 ROS 节点
    rospy.init_node("static_tf_pub_p")
    # 3.创建 静态坐标广播器
    broadcaster = tf2_ros.StaticTransformBroadcaster()
    # 4.创建并组织被广播的消息
    tfs = TransformStamped()
    # --- 头信息
    tfs.header.frame_id = "world"
    tfs.header.stamp = rospy.Time.now()
    tfs.header.seq = 101
    # --- 子坐标系
    tfs.child_frame_id = "radar"
    # --- 坐标系相对信息
    # ------ 偏移量
    tfs.transform.translation.x = 0.2
    tfs.transform.translation.y = 0.0
    tfs.transform.translation.z = 0.5
    # ------ 四元数
    qtn = tf.transformations.quaternion_from_euler(0,0,0)
    tfs.transform.rotation.x = qtn[0]
    tfs.transform.rotation.y = qtn[1]
    tfs.transform.rotation.z = qtn[2]
    tfs.transform.rotation.w = qtn[3]


    # 5.广播器发送消息
    broadcaster.sendTransform(tfs)
    # 6.spin
    rospy.spin()

Permission settings and configuration files are omitted here.

Example result:

3. Subscriber

#! /usr/bin/env python
"""  
    订阅坐标系信息，生成一个相对于 子级坐标系的坐标点数据，
    转换成父级坐标系中的坐标点

    实现流程:
        1.导包
        2.初始化 ROS 节点
        3.创建 TF 订阅对象
        4.创建一个 radar 坐标系中的坐标点
        5.调研订阅对象的 API 将 4 中的点坐标转换成相对于 world 的坐标
        6.spin

"""
# 1.导包
import rospy
import tf2_ros
# 不要使用 geometry_msgs,需要使用 tf2 内置的消息类型
from tf2_geometry_msgs import PointStamped
# from geometry_msgs.msg import PointStamped

if __name__ == "__main__":
    # 2.初始化 ROS 节点
    rospy.init_node("static_sub_tf_p")
    # 3.创建 TF 订阅对象
    buffer = tf2_ros.Buffer()
    listener = tf2_ros.TransformListener(buffer)

    rate = rospy.Rate(1)
    while not rospy.is_shutdown():    
    # 4.创建一个 radar 坐标系中的坐标点
        point_source = PointStamped()
        point_source.header.frame_id = "radar"
        point_source.header.stamp = rospy.Time.now()
        point_source.point.x = 10
        point_source.point.y = 2
        point_source.point.z = 3

        try:
    #     5.调研订阅对象的 API 将 4 中的点坐标转换成相对于 world 的坐标
            point_target = buffer.transform(point_source,"world")
            rospy.loginfo("转换结果:x = %.2f, y = %.2f, z = %.2f",
                            point_target.point.x,
                            point_target.point.y,
                            point_target.point.z)
        except Exception as e:
            rospy.logerr("异常:%s",e)

    #     6.spin
        rate.sleep()

Permission settings and configuration files are omitted here.

PS: In the python implementation of tf2, tf2 has encapsulated some message types, and the types in geometry_msgs.msg cannot be used

4. Execution

You can use the command line or the launch file to start the publishing node and the subscribing node respectively. If there is no abnormality in the program, the console will output the coordinate conversion result.

Example result:

Supplement 1:

When the relative position between coordinate systems is fixed, the required parameters are also fixed: parent coordinate name, child coordinate system name, x offset, y offset, z offset, x roll angle, y pitch Angle, z yaw angle, the implementation logic is the same, but the parameters are different, then the ROS system has already packaged a special node, and the usage method is as follows:

rosrun tf2_ros static_transform_publisher x偏移量 y偏移量 z偏移量 z偏航角度 y俯仰角度 x翻滚角度 父级坐标系 子级坐标系

Example:rosrun tf2_ros static_transform_publisher 0.2 0 0.5 0 0 0 /baselink /laser

It is also recommended to use this method to directly realize the relative information release of the static coordinate system.

Supplement 2:

You can use rviz to display the coordinate system relationship, the specific operation:

• Enter the command in the new window: rviz;
• Set Fixed Frame to base_link in the started rviz;
• Click the add button on the lower left, and select the TF component in the pop-up window to display the coordinate relationship.

---

2. Dynamic coordinate transformation

The so-called dynamic coordinate transformation means that the relative position between two coordinate systems changes.

Description of Requirement:

Start turtlesim_node, the window in this node has a world coordinate system (the lower left corner is the origin of the coordinate system), and the turtle is another coordinate system. The keyboard controls the movement of the turtle, and dynamically publishes the relative positions of the two coordinate systems.

Result demo:

Realize analysis:

1. The turtle itself can not only be regarded as a coordinate system, but also a coordinate point in the world coordinate system
2. Subscribe to turtle1/pose, you can get the x-coordinate, y-coordinate, offset, linear velocity and angular velocity of the turtle in the world coordinate system
3. Convert pose information into coordinate system relative information and release

Implementation process: The implementation process of C++ and Python is consistent

1. Create a new feature package and add dependencies
2. Create a coordinate relative relationship publisher (need to subscribe to turtle pose information at the same time)
3. Create a coordinate relative relationship subscriber
4. implement

---

Solution A: C++ implementation

1. Create a feature package

Create a project feature package that depends on

 tf2、tf2_ros、tf2_geometry_msgs、roscpp rospy std_msgs geometry_msgs、turtlesim

2. Publisher

/*  
    动态的坐标系相对姿态发布(一个坐标系相对于另一个坐标系的相对姿态是不断变动的)

    需求: 启动 turtlesim_node,该节点中窗体有一个世界坐标系(左下角为坐标系原点)，乌龟是另一个坐标系，键盘
    控制乌龟运动，将两个坐标系的相对位置动态发布

    实现分析:
        1.乌龟本身不但可以看作坐标系，也是世界坐标系中的一个坐标点
        2.订阅 turtle1/pose,可以获取乌龟在世界坐标系的 x坐标、y坐标、偏移量以及线速度和角速度
        3.将 pose 信息转换成 坐标系相对信息并发布

    实现流程:
        1.包含头文件
        2.初始化 ROS 节点
        3.创建 ROS 句柄
        4.创建订阅对象
        5.回调函数处理订阅到的数据(实现TF广播)
            5-1.创建 TF 广播器
            5-2.创建 广播的数据(通过 pose 设置)
            5-3.广播器发布数据
        6.spin
*/
// 1.包含头文件
#include "ros/ros.h"
#include "turtlesim/Pose.h"
#include "tf2_ros/transform_broadcaster.h"
#include "geometry_msgs/TransformStamped.h"
#include "tf2/LinearMath/Quaternion.h"

void doPose(const turtlesim::Pose::ConstPtr& pose){
    
    
    //  5-1.创建 TF 广播器
    static tf2_ros::TransformBroadcaster broadcaster;
    //  5-2.创建 广播的数据(通过 pose 设置)
    geometry_msgs::TransformStamped tfs;
    //  |----头设置
    tfs.header.frame_id = "world";
    tfs.header.stamp = ros::Time::now();

    //  |----坐标系 ID
    tfs.child_frame_id = "turtle1";

    //  |----坐标系相对信息设置
    tfs.transform.translation.x = pose->x;
    tfs.transform.translation.y = pose->y;
    tfs.transform.translation.z = 0.0; // 二维实现，pose 中没有z，z 是 0
    //  |--------- 四元数设置
    tf2::Quaternion qtn;
    qtn.setRPY(0,0,pose->theta);
    tfs.transform.rotation.x = qtn.getX();
    tfs.transform.rotation.y = qtn.getY();
    tfs.transform.rotation.z = qtn.getZ();
    tfs.transform.rotation.w = qtn.getW();


    //  5-3.广播器发布数据
    broadcaster.sendTransform(tfs);
}

int main(int argc, char *argv[])
{
    
    
    setlocale(LC_ALL,"");
    // 2.初始化 ROS 节点
    ros::init(argc,argv,"dynamic_tf_pub");
    // 3.创建 ROS 句柄
    ros::NodeHandle nh;
    // 4.创建订阅对象
    ros::Subscriber sub = nh.subscribe<turtlesim::Pose>("/turtle1/pose",1000,doPose);
    //     5.回调函数处理订阅到的数据(实现TF广播)
    //        
    // 6.spin
    ros::spin();
    return 0;
}

Realize the effect:

The configuration file is omitted here.

3. Subscriber

//1.包含头文件
#include "ros/ros.h"
#include "tf2_ros/transform_listener.h"
#include "tf2_ros/buffer.h"
#include "geometry_msgs/PointStamped.h"
#include "tf2_geometry_msgs/tf2_geometry_msgs.h" //注意: 调用 transform 必须包含该头文件

int main(int argc, char *argv[])
{
    
    
    setlocale(LC_ALL,"");
    // 2.初始化 ROS 节点
    ros::init(argc,argv,"dynamic_tf_sub");
    ros::NodeHandle nh;
    // 3.创建 TF 订阅节点
    tf2_ros::Buffer buffer;
    tf2_ros::TransformListener listener(buffer);

    ros::Rate r(1);
    while (ros::ok())
    {
    
    
    // 4.生成一个坐标点(相对于子级坐标系)
        geometry_msgs::PointStamped point_laser;
        point_laser.header.frame_id = "turtle1";
        point_laser.header.stamp = ros::Time();
        point_laser.point.x = 1;
        point_laser.point.y = 1;
        point_laser.point.z = 0;
    // 5.转换坐标点(相对于父级坐标系)
        //新建一个坐标点，用于接收转换结果  
        //--------------使用 try 语句或休眠，否则可能由于缓存接收延迟而导致坐标转换失败------------------------
        try
        {
    
    
            geometry_msgs::PointStamped point_base;
            point_base = buffer.transform(point_laser,"world");
            ROS_INFO("坐标点相对于 world 的坐标为:(%.2f,%.2f,%.2f)",point_base.point.x,point_base.point.y,point_base.point.z);

        }
        catch(const std::exception& e)
        {
    
    
            // std::cerr << e.what() << '\n';
            ROS_INFO("程序异常:%s",e.what());
        }


        r.sleep();  
        ros::spinOnce();
    }


    return 0;
}

The configuration file is omitted here.

4. Execution

You can use the command line or launch file to start the publishing node and the subscribing node respectively. If the program is normal, the result is similar to the demonstration.

You can use rviz to view the relative relationship of the coordinate system.

Example result:

---

Option B: Python implementation

1. Create a feature package

Create a project feature package that depends on

 tf2、tf2_ros、tf2_geometry_msgs、roscpp rospy std_msgs geometry_msgs、turtlesim

2. Publisher

#! /usr/bin/env python
"""  
    动态的坐标系相对姿态发布(一个坐标系相对于另一个坐标系的相对姿态是不断变动的)

    需求: 启动 turtlesim_node,该节点中窗体有一个世界坐标系(左下角为坐标系原点)，乌龟是另一个坐标系，键盘
    控制乌龟运动，将两个坐标系的相对位置动态发布

    实现分析:
        1.乌龟本身不但可以看作坐标系，也是世界坐标系中的一个坐标点
        2.订阅 turtle1/pose,可以获取乌龟在世界坐标系的 x坐标、y坐标、偏移量以及线速度和角速度
        3.将 pose 信息转换成 坐标系相对信息并发布
    实现流程:
        1.导包
        2.初始化 ROS 节点
        3.订阅 /turtle1/pose 话题消息
        4.回调函数处理
            4-1.创建 TF 广播器
            4-2.创建 广播的数据(通过 pose 设置)
            4-3.广播器发布数据
        5.spin
"""
# 1.导包
import rospy
import tf2_ros
import tf
from turtlesim.msg import Pose
from geometry_msgs.msg import TransformStamped

#     4.回调函数处理
def doPose(pose):
    #         4-1.创建 TF 广播器
    broadcaster = tf2_ros.TransformBroadcaster()
    #         4-2.创建 广播的数据(通过 pose 设置)
    tfs = TransformStamped()
    tfs.header.frame_id = "world"
    tfs.header.stamp = rospy.Time.now()
    tfs.child_frame_id = "turtle1"
    tfs.transform.translation.x = pose.x
    tfs.transform.translation.y = pose.y
    tfs.transform.translation.z = 0.0
    qtn = tf.transformations.quaternion_from_euler(0,0,pose.theta)
    tfs.transform.rotation.x = qtn[0]
    tfs.transform.rotation.y = qtn[1]
    tfs.transform.rotation.z = qtn[2]
    tfs.transform.rotation.w = qtn[3]
    #         4-3.广播器发布数据
    broadcaster.sendTransform(tfs)

if __name__ == "__main__":
    # 2.初始化 ROS 节点
    rospy.init_node("dynamic_tf_pub_p")
    # 3.订阅 /turtle1/pose 话题消息
    sub = rospy.Subscriber("/turtle1/pose",Pose,doPose)
    #     4.回调函数处理
    #         4-1.创建 TF 广播器
    #         4-2.创建 广播的数据(通过 pose 设置)
    #         4-3.广播器发布数据
    #     5.spin
    rospy.spin()

Permission settings and configuration files are omitted here.

The effect of the example is as follows:

3. Subscriber

#! /usr/bin/env python
"""  
    动态的坐标系相对姿态发布(一个坐标系相对于另一个坐标系的相对姿态是不断变动的)

    需求: 启动 turtlesim_node,该节点中窗体有一个世界坐标系(左下角为坐标系原点)，乌龟是另一个坐标系，键盘
    控制乌龟运动，将两个坐标系的相对位置动态发布

    实现分析:
        1.乌龟本身不但可以看作坐标系，也是世界坐标系中的一个坐标点
        2.订阅 turtle1/pose,可以获取乌龟在世界坐标系的 x坐标、y坐标、偏移量以及线速度和角速度
        3.将 pose 信息转换成 坐标系相对信息并发布
    实现流程:
        1.导包
        2.初始化 ROS 节点
        3.创建 TF 订阅对象
        4.处理订阅的数据


"""
# 1.导包
import rospy
import tf2_ros
# 不要使用 geometry_msgs,需要使用 tf2 内置的消息类型
from tf2_geometry_msgs import PointStamped
# from geometry_msgs.msg import PointStamped

if __name__ == "__main__":
    # 2.初始化 ROS 节点
    rospy.init_node("static_sub_tf_p")
    # 3.创建 TF 订阅对象
    buffer = tf2_ros.Buffer()
    listener = tf2_ros.TransformListener(buffer)

    rate = rospy.Rate(1)
    while not rospy.is_shutdown():    
    # 4.创建一个 radar 坐标系中的坐标点
        point_source = PointStamped()
        point_source.header.frame_id = "turtle1"
        point_source.header.stamp = rospy.Time.now()
        point_source.point.x = 10
        point_source.point.y = 2
        point_source.point.z = 3

        try:
    #     5.调研订阅对象的 API 将 4 中的点坐标转换成相对于 world 的坐标
            point_target = buffer.transform(point_source,"world",rospy.Duration(1))
            rospy.loginfo("转换结果:x = %.2f, y = %.2f, z = %.2f",
                            point_target.point.x,
                            point_target.point.y,
                            point_target.point.z)
        except Exception as e:
            rospy.logerr("异常:%s",e)

    #     6.spin
        rate.sleep()

Permission settings and configuration files are omitted here.

4. Execution

You can use the command line or launch file to start the publishing node and the subscribing node respectively. If the program is normal, the result is similar to the demonstration.

You can use rviz to view the relative relationship of the coordinate system.

Example results are as follows:

---

3. Multi-coordinate transformation

Description of Requirement:

The existing coordinate system, the parent coordinate system world, has two sub-systems son1, son2, the relationship between son1 relative to the world, and son2 relative to the world is known, and the coordinates of the origin of son1 in son2 are known. The coordinates of a point in son1 are required to find the coordinates of the point in son2

Realize analysis:

1. First, you need to publish the coordinate messages of son1 relative to the world and son2 relative to the world
2. Then, you need to subscribe to the coordinates to publish messages, and take out the subscribed messages, and use tf2 to realize the conversion between son1 and son2
3. Finally, it is necessary to realize the conversion of coordinate points

**Implementation process:**C++ is consistent with Python implementation process

1. Create a new feature package and add dependencies
2. Create a coordinate relative relationship publisher (need to publish two coordinate relative relationships)
3. Create a coordinate relative relationship subscriber
4. implement

Solution A: C++ implementation

1. Create a feature package

Create project function package depends on tf2, tf2_ros, tf2_geometry_msgs, roscpp rospy std_msgs geometry_msgs, turtlesim

2. Publisher

For convenience, publish using static coordinate transformation

<launch>
    <node pkg="tf2_ros" type="static_transform_publisher" name="son1" args="5 0 0 0 0 0 /world /son1" output="screen" />
    <node pkg="tf2_ros" type="static_transform_publisher" name="son2" args="3 0 0 0 0 0 /world /son2" output="screen" />
</launch>

Example result:

3. Subscriber

/*

需求:
    现有坐标系统，父级坐标系统 world,下有两子级系统 son1，son2，
    son1 相对于 world，以及 son2 相对于 world 的关系是已知的，
    求 son1 与 son2中的坐标关系，又已知在 son1中一点的坐标，要求求出该点在 son2 中的坐标
实现流程:
    1.包含头文件
    2.初始化 ros 节点
    3.创建 ros 句柄
    4.创建 TF 订阅对象
    5.解析订阅信息中获取 son1 坐标系原点在 son2 中的坐标
      解析 son1 中的点相对于 son2 的坐标
    6.spin

*/
//1.包含头文件
#include "ros/ros.h"
#include "tf2_ros/transform_listener.h"
#include "tf2_ros/buffer.h"
#include "tf2_geometry_msgs/tf2_geometry_msgs.h"
#include "geometry_msgs/TransformStamped.h"
#include "geometry_msgs/PointStamped.h"

int main(int argc, char *argv[])
{
    
       setlocale(LC_ALL,"");
    // 2.初始化 ros 节点
    ros::init(argc,argv,"sub_frames");
    // 3.创建 ros 句柄
    ros::NodeHandle nh;
    // 4.创建 TF 订阅对象
    tf2_ros::Buffer buffer; 
    tf2_ros::TransformListener listener(buffer);
    // 5.解析订阅信息中获取 son1 坐标系原点在 son2 中的坐标
    ros::Rate r(1);
    while (ros::ok())
    {
    
    
        try
        {
    
    
        //   解析 son1 中的点相对于 son2 的坐标
            geometry_msgs::TransformStamped tfs = buffer.lookupTransform("son2","son1",ros::Time(0));
            ROS_INFO("Son1 相对于 Son2 的坐标关系:父坐标系ID=%s",tfs.header.frame_id.c_str());
            ROS_INFO("Son1 相对于 Son2 的坐标关系:子坐标系ID=%s",tfs.child_frame_id.c_str());
            ROS_INFO("Son1 相对于 Son2 的坐标关系:x=%.2f,y=%.2f,z=%.2f",
                    tfs.transform.translation.x,
                    tfs.transform.translation.y,
                    tfs.transform.translation.z
                    );

            // 坐标点解析
            geometry_msgs::PointStamped ps;
            ps.header.frame_id = "son1";
            ps.header.stamp = ros::Time::now();
            ps.point.x = 1.0;
            ps.point.y = 2.0;
            ps.point.z = 3.0;

            geometry_msgs::PointStamped psAtSon2;
            psAtSon2 = buffer.transform(ps,"son2");
            ROS_INFO("在 Son2 中的坐标:x=%.2f,y=%.2f,z=%.2f",
                    psAtSon2.point.x,
                    psAtSon2.point.y,
                    psAtSon2.point.z
                    );
        }
        catch(const std::exception& e)
        {
    
    
            // std::cerr << e.what() << '\n';
            ROS_INFO("异常信息:%s",e.what());
        }


        r.sleep();
        // 6.spin
        ros::spinOnce();
    }
    return 0;
}

The configuration file is omitted here.

4. Execution

You can use the command line or launch file to start the publishing node and the subscribing node respectively. If the program is normal, the converted result will be output.

Example result:

Option B: Python implementation

1. Create a feature package

Create project function package depends on tf2, tf2_ros, tf2_geometry_msgs, roscpp rospy std_msgs geometry_msgs, turtlesim

2. Publisher

For convenience, publish using static coordinate transformation

<launch>
    <node pkg="tf2_ros" type="static_transform_publisher" name="son1" args="5 0 0 0 0 0 /world /son1" output="screen" />
    <node pkg="tf2_ros" type="static_transform_publisher" name="son2" args="3 0 0 0 0 0 /world /son2" output="screen" />
</launch>

3. Subscriber

#!/usr/bin/env python
"""  
    需求:
        现有坐标系统，父级坐标系统 world,下有两子级系统 son1，son2，
        son1 相对于 world，以及 son2 相对于 world 的关系是已知的，
        求 son1 与 son2中的坐标关系，又已知在 son1中一点的坐标，要求求出该点在 son2 中的坐标
    实现流程:   
        1.导包
        2.初始化 ROS 节点
        3.创建 TF 订阅对象
        4.调用 API 求出 son1 相对于 son2 的坐标关系
        5.创建一依赖于 son1 的坐标点，调用 API 求出该点在 son2 中的坐标
        6.spin

"""
# 1.导包
import rospy
import tf2_ros
from geometry_msgs.msg import TransformStamped
from tf2_geometry_msgs import PointStamped

if __name__ == "__main__":

    # 2.初始化 ROS 节点
    rospy.init_node("frames_sub_p")
    # 3.创建 TF 订阅对象
    buffer = tf2_ros.Buffer()
    listener = tf2_ros.TransformListener(buffer)

    rate = rospy.Rate(1)
    while not rospy.is_shutdown():

        try:
        # 4.调用 API 求出 son1 相对于 son2 的坐标关系
            #lookup_transform(self, target_frame, source_frame, time, timeout=rospy.Duration(0.0)):
            tfs = buffer.lookup_transform("son2","son1",rospy.Time(0))
            rospy.loginfo("son1 与 son2 相对关系:")
            rospy.loginfo("父级坐标系:%s",tfs.header.frame_id)
            rospy.loginfo("子级坐标系:%s",tfs.child_frame_id)
            rospy.loginfo("相对坐标:x=%.2f, y=%.2f, z=%.2f",
                        tfs.transform.translation.x,
                        tfs.transform.translation.y,
                        tfs.transform.translation.z,
            )
        # 5.创建一依赖于 son1 的坐标点，调用 API 求出该点在 son2 中的坐标
            point_source = PointStamped()
            point_source.header.frame_id = "son1"
            point_source.header.stamp = rospy.Time.now()
            point_source.point.x = 1
            point_source.point.y = 1
            point_source.point.z = 1

            point_target = buffer.transform(point_source,"son2",rospy.Duration(0.5))

            rospy.loginfo("point_target 所属的坐标系:%s",point_target.header.frame_id)
            rospy.loginfo("坐标点相对于 son2 的坐标:(%.2f,%.2f,%.2f)",
                        point_target.point.x,
                        point_target.point.y,
                        point_target.point.z
            )

        except Exception as e:
            rospy.logerr("错误提示:%s",e)


        rate.sleep()
    # 6.spin    
    # rospy.spin()

Permission settings and configuration files are omitted here.

4. Execution

You can use the command line or launch file to start the publishing node and the subscribing node respectively. If the program is normal, the converted result will be output.

---

Example result:

---

4. View the coordinate system relationship

In the robot system, there are multiple coordinate systems involved. For the convenience of viewing, ros provides a special tool that can be used to generate a pdf file showing the coordinate system relationship, which contains a tree-structured coordinate system map.

1. Prepare

First call rospack find tf2_toolsto check whether the feature package is included, if not, please use the following command to install:

sudo apt install ros-noetic-tf2-tools

2. use

After starting the coordinate system broadcast program, run the following command to generate a pdf file

rosrun tf2_tools view_frames.py

A log message similar to the following will be generated:

[INFO] [1592920556.827549]: Listening to tf data during 5 seconds...
[INFO] [1592920561.841536]: Generating graph in frames.pdf file...

Viewing the current directory will generate a frames.pdf file

You can directly enter the directory to open the file, or call the command to view the file: evince frames.pdf, to view the file

Inside as shown in the figure:

---

5. TF coordinate transformation practice

Description of Requirement:

At the beginning of the program: Two turtles are generated, the middle turtle (A) and the lower left turtle (B), B will automatically run to the position of A, and when the keyboard is controlled, it only controls the movement of A, but B can follow A to run

Result demo:

Realize analysis:

The core of the turtle following implementation is that both turtles A and B publish coordinate information relative to the world coordinate system. Then, to subscribe to this information, you need to convert and obtain the information of A relative to the coordinate system of B. Finally, generate speed information and control B. sports.

1. Start the turtle display node
2. Spawns a new turtle in the turtle display form (requires service)
3. Write a node for two turtles to publish coordinate information
4. Write a subscription node to subscribe to coordinate information and generate a new relative relationship to generate speed information

**Implementation process:**C++ is consistent with Python implementation process

1. Create a new feature package and add dependencies
2. Write a service client for spawning a new turtle
3. Write the publisher to publish the coordinate information of the two turtles
4. Write a subscriber, subscribe to the information of two turtles, generate speed information and publish it
5. run

Preparation:

1. Learn how to create a second turtle that is not controlled by the keyboard

To create a second turtle, you need to use rosservice, and the topic uses spawn

rosservice call /spawn "x: 1.0
y: 1.0
theta: 1.0
name: 'turtle_flow'" 
name: "turtle_flow"

The keyboard cannot control the movement of the second turtle, because the topic used: /name of the second turtle/cmd_vel, the corresponding topic message must be published to control the movement of the turtle

2. Learn how to get the coordinates of two turtles

is 话题 /乌龟名称/pose obtained through

x: 1.0 //x坐标
y: 1.0 //y坐标
theta: -1.21437060833 //角度
linear_velocity: 0.0 //线速度
angular_velocity: 1.0 //角速度

Solution A: C++ implementation

1. Create a feature package

Create a project feature package that depends on

tf2、tf2_ros、tf2_geometry_msgs、roscpp rospy std_msgs geometry_msgs、turtlesim

2. Service client (generate turtle)

/* 
    创建第二只小乌龟
 */
#include "ros/ros.h"
#include "turtlesim/Spawn.h"

int main(int argc, char *argv[])
{
    
    

    setlocale(LC_ALL,"");

    //执行初始化
    ros::init(argc,argv,"create_turtle");
    //创建节点
    ros::NodeHandle nh;
    //创建服务客户端
    ros::ServiceClient client = nh.serviceClient<turtlesim::Spawn>("/spawn");

    ros::service::waitForService("/spawn");
    turtlesim::Spawn spawn;
    spawn.request.name = "turtle2";
    spawn.request.x = 1.0;
    spawn.request.y = 2.0;
    spawn.request.theta = 3.12415926;
    bool flag = client.call(spawn);
    if (flag)
    {
    
    
        ROS_INFO("乌龟%s创建成功!",spawn.response.name.c_str());
    }
    else
    {
    
    
        ROS_INFO("乌龟2创建失败!");
    }

    ros::spin();

    return 0;
}

The configuration file is omitted here.

3. The publisher (publishes the coordinate information of the two turtles)

You can subscribe to the pose information of the turtle, and then convert it into coordinate information. The implementation logic of the two turtles is the same, but the subscribed topic name and the generated coordinate information are slightly different. The difference can be passed in through parameters:

• The node needs to be started twice
• You need to pass in the turtle node name every time you start it (the first time is turtle1 and the second time is turtle2)

/*  
    该文件实现:需要订阅 turtle1 和 turtle2 的 pose，然后广播相对 world 的坐标系信息

    注意: 订阅的两只 turtle,除了命名空间(turtle1 和 turtle2)不同外,
          其他的话题名称和实现逻辑都是一样的，
          所以我们可以将所需的命名空间通过 args 动态传入

    实现流程:
        1.包含头文件
        2.初始化 ros 节点
        3.解析传入的命名空间
        4.创建 ros 句柄
        5.创建订阅对象
        6.回调函数处理订阅的 pose 信息
            6-1.创建 TF 广播器
            6-2.将 pose 信息转换成 TransFormStamped
            6-3.发布
        7.spin

*/
//1.包含头文件
#include "ros/ros.h"
#include "turtlesim/Pose.h"
#include "tf2_ros/transform_broadcaster.h"
#include "tf2/LinearMath/Quaternion.h"
#include "geometry_msgs/TransformStamped.h"
//保存乌龟名称
std::string turtle_name;


void doPose(const turtlesim::Pose::ConstPtr& pose){
    
    
    //  6-1.创建 TF 广播器 ---------------------------------------- 注意 static
    static tf2_ros::TransformBroadcaster broadcaster;
    //  6-2.将 pose 信息转换成 TransFormStamped
    geometry_msgs::TransformStamped tfs;
    tfs.header.frame_id = "world";
    tfs.header.stamp = ros::Time::now();
    tfs.child_frame_id = turtle_name;
    tfs.transform.translation.x = pose->x;
    tfs.transform.translation.y = pose->y;
    tfs.transform.translation.z = 0.0;
    tf2::Quaternion qtn;
    qtn.setRPY(0,0,pose->theta);
    tfs.transform.rotation.x = qtn.getX();
    tfs.transform.rotation.y = qtn.getY();
    tfs.transform.rotation.z = qtn.getZ();
    tfs.transform.rotation.w = qtn.getW();
    //  6-3.发布
    broadcaster.sendTransform(tfs);

} 

int main(int argc, char *argv[])
{
    
    
    setlocale(LC_ALL,"");
    // 2.初始化 ros 节点
    ros::init(argc,argv,"pub_tf");
    // 3.解析传入的命名空间
    if (argc != 2)
    {
    
    
        ROS_ERROR("请传入正确的参数");
    } else {
    
    
        turtle_name = argv[1];
        ROS_INFO("乌龟 %s 坐标发送启动",turtle_name.c_str());
    }

    // 4.创建 ros 句柄
    ros::NodeHandle nh;
    // 5.创建订阅对象
    ros::Subscriber sub = nh.subscribe<turtlesim::Pose>(turtle_name + "/pose",1000,doPose);
    //     6.回调函数处理订阅的 pose 信息
    //         6-1.创建 TF 广播器
    //         6-2.将 pose 信息转换成 TransFormStamped
    //         6-3.发布
    // 7.spin
    ros::spin();
    return 0;
}

The configuration file is omitted here.

4. Subscriber (parse coordinate information and generate speed information)

/*  
    订阅 turtle1 和 turtle2 的 TF 广播信息，查找并转换时间最近的 TF 信息
    将 turtle1 转换成相对 turtle2 的坐标，在计算线速度和角速度并发布

    实现流程:
        1.包含头文件
        2.初始化 ros 节点
        3.创建 ros 句柄
        4.创建 TF 订阅对象
        5.处理订阅到的 TF
        6.spin

*/
//1.包含头文件
#include "ros/ros.h"
#include "tf2_ros/transform_listener.h"
#include "geometry_msgs/TransformStamped.h"
#include "geometry_msgs/Twist.h"

int main(int argc, char *argv[])
{
    
    
    setlocale(LC_ALL,"");
    // 2.初始化 ros 节点
    ros::init(argc,argv,"sub_TF");
    // 3.创建 ros 句柄
    ros::NodeHandle nh;
    // 4.创建 TF 订阅对象
    tf2_ros::Buffer buffer;
    tf2_ros::TransformListener listener(buffer);
    // 5.处理订阅到的 TF

    // 需要创建发布 /turtle2/cmd_vel 的 publisher 对象

    ros::Publisher pub = nh.advertise<geometry_msgs::Twist>("/turtle2/cmd_vel",1000);

    ros::Rate rate(10);
    while (ros::ok())
    {
    
    
        try
        {
    
    
            //5-1.先获取 turtle1 相对 turtle2 的坐标信息
            geometry_msgs::TransformStamped tfs = buffer.lookupTransform("turtle2","turtle1",ros::Time(0));

            //5-2.根据坐标信息生成速度信息 -- geometry_msgs/Twist.h
            geometry_msgs::Twist twist;
            twist.linear.x = 0.5 * sqrt(pow(tfs.transform.translation.x,2) + pow(tfs.transform.translation.y,2));
            twist.angular.z = 4 * atan2(tfs.transform.translation.y,tfs.transform.translation.x);

            //5-3.发布速度信息 -- 需要提前创建 publish 对象
            pub.publish(twist);
        }
        catch(const std::exception& e)
        {
    
    
            // std::cerr << e.what() << '\n';
            ROS_INFO("错误提示:%s",e.what());
        }



        rate.sleep();
        // 6.spin
        ros::spinOnce();
    }

    return 0;
}

The configuration file is omitted here.

5. run

Use the launch file to organize the nodes that need to be run, the content example is as follows:

<!--
    tf2 实现小乌龟跟随案例
-->
<launch>
    <!-- 启动乌龟节点与键盘控制节点 -->
    <node pkg="turtlesim" type="turtlesim_node" name="turtle1" output="screen" />
    <node pkg="turtlesim" type="turtle_teleop_key" name="key_control" output="screen"/>
    <!-- 启动创建第二只乌龟的节点 -->
    <node pkg="demo_tf2_test" type="Test01_Create_Turtle2" name="turtle2" output="screen" />
    <!-- 启动两个坐标发布节点 -->
    <node pkg="demo_tf2_test" type="Test02_TF2_Caster" name="caster1" output="screen" args="turtle1" />
    <node pkg="demo_tf2_test" type="Test02_TF2_Caster" name="caster2" output="screen" args="turtle2" />
    <!-- 启动坐标转换节点 -->
    <node pkg="demo_tf2_test" type="Test03_TF2_Listener" name="listener" output="screen" />
</launch>

Option B: Python implementation

1. Create a feature package

Create a project feature package that depends on

tf2、tf2_ros、tf2_geometry_msgs、roscpp rospy std_msgs geometry_msgs、turtlesim

2. Service client (generate turtle)

#! /usr/bin/env python
"""  
    调用 service 服务在窗体指定位置生成一只乌龟
    流程:
        1.导包
        2.初始化 ros 节点
        3.创建服务客户端
        4.等待服务启动
        5.创建请求数据
        6.发送请求并处理响应
"""
#1.导包
import rospy
from turtlesim.srv import Spawn, SpawnRequest, SpawnResponse

if __name__ == "__main__":
    # 2.初始化 ros 节点
    rospy.init_node("turtle_spawn_p")
    # 3.创建服务客户端
    client = rospy.ServiceProxy("/spawn",Spawn)
    # 4.等待服务启动
    client.wait_for_service()
    # 5.创建请求数据
    req = SpawnRequest()
    req.x = 1.0
    req.y = 1.0
    req.theta = 3.14
    req.name = "turtle2"
    # 6.发送请求并处理响应
    try:
        response = client.call(req)
        rospy.loginfo("乌龟创建成功，名字是:%s",response.name)
    except Exception as e:
        rospy.loginfo("服务调用失败....")

Permission settings and configuration files are omitted here.

3. The publisher (publishes the coordinate information of the two turtles)

#! /usr/bin/env python
"""  
    该文件实现:需要订阅 turtle1 和 turtle2 的 pose，然后广播相对 world 的坐标系信息

    注意: 订阅的两只 turtle,除了命名空间(turtle1 和 turtle2)不同外,
          其他的话题名称和实现逻辑都是一样的，
          所以我们可以将所需的命名空间通过 args 动态传入

    实现流程:
        1.导包
        2.初始化 ros 节点
        3.解析传入的命名空间
        4.创建订阅对象
        5.回调函数处理订阅的 pose 信息
            5-1.创建 TF 广播器
            5-2.将 pose 信息转换成 TransFormStamped
            5-3.发布
        6.spin
"""
# 1.导包
import rospy
import sys
from turtlesim.msg import Pose
from geometry_msgs.msg import TransformStamped
import tf2_ros
import tf_conversions

turtle_name = ""

def doPose(pose):
    # rospy.loginfo("x = %.2f",pose.x)
    #1.创建坐标系广播器
    broadcaster = tf2_ros.TransformBroadcaster()
    #2.将 pose 信息转换成 TransFormStamped
    tfs = TransformStamped()
    tfs.header.frame_id = "world"
    tfs.header.stamp = rospy.Time.now()

    tfs.child_frame_id = turtle_name
    tfs.transform.translation.x = pose.x
    tfs.transform.translation.y = pose.y
    tfs.transform.translation.z = 0.0

    qtn = tf_conversions.transformations.quaternion_from_euler(0, 0, pose.theta)
    tfs.transform.rotation.x = qtn[0]
    tfs.transform.rotation.y = qtn[1]
    tfs.transform.rotation.z = qtn[2]
    tfs.transform.rotation.w = qtn[3]

    #3.广播器发布 tfs
    broadcaster.sendTransform(tfs)


if __name__ == "__main__":
    # 2.初始化 ros 节点
    rospy.init_node("sub_tfs_p")
    # 3.解析传入的命名空间
    rospy.loginfo("-------------------------------%d",len(sys.argv))
    if len(sys.argv) < 2:
        rospy.loginfo("请传入参数:乌龟的命名空间")
    else:
        turtle_name = sys.argv[1]
    rospy.loginfo("///乌龟:%s",turtle_name)

    rospy.Subscriber(turtle_name + "/pose",Pose,doPose)
    #     4.创建订阅对象
    #     5.回调函数处理订阅的 pose 信息
    #         5-1.创建 TF 广播器
    #         5-2.将 pose 信息转换成 TransFormStamped
    #         5-3.发布
    #     6.spin
    rospy.spin()

Permission settings and configuration files are omitted here.

4. Subscriber (parse coordinate information and generate speed information)

#! /usr/bin/env python
"""  
    订阅 turtle1 和 turtle2 的 TF 广播信息，查找并转换时间最近的 TF 信息
    将 turtle1 转换成相对 turtle2 的坐标，在计算线速度和角速度并发布

    实现流程:
        1.导包
        2.初始化 ros 节点
        3.创建 TF 订阅对象
        4.处理订阅到的 TF
            4-1.查找坐标系的相对关系
            4-2.生成速度信息，然后发布
"""
# 1.导包
import rospy
import tf2_ros
from geometry_msgs.msg import TransformStamped, Twist
import math

if __name__ == "__main__":
    # 2.初始化 ros 节点
    rospy.init_node("sub_tfs_p")
    # 3.创建 TF 订阅对象
    buffer = tf2_ros.Buffer()
    listener = tf2_ros.TransformListener(buffer)
    # 4.处理订阅到的 TF
    rate = rospy.Rate(10)
    # 创建速度发布对象
    pub = rospy.Publisher("/turtle2/cmd_vel",Twist,queue_size=1000)
    while not rospy.is_shutdown():

        rate.sleep()
        try:
            #def lookup_transform(self, target_frame, source_frame, time, timeout=rospy.Duration(0.0)):
            trans = buffer.lookup_transform("turtle2","turtle1",rospy.Time(0))
            # rospy.loginfo("相对坐标:(%.2f,%.2f,%.2f)",
            #             trans.transform.translation.x,
            #             trans.transform.translation.y,
            #             trans.transform.translation.z
            #             )   
            # 根据转变后的坐标计算出速度和角速度信息
            twist = Twist()
            # 间距 = x^2 + y^2  然后开方
            twist.linear.x = 0.5 * math.sqrt(math.pow(trans.transform.translation.x,2) + math.pow(trans.transform.translation.y,2))
            twist.angular.z = 4 * math.atan2(trans.transform.translation.y, trans.transform.translation.x)

            pub.publish(twist)

        except Exception as e:
            rospy.logwarn("警告:%s",e)

Permission settings and configuration files are omitted here.

5. run

Use the launch file to organize the nodes that need to be run, the content example is as follows:

<launch>
    <node pkg="turtlesim" type="turtlesim_node" name="turtle1" output="screen" />
    <node pkg="turtlesim" type="turtle_teleop_key" name="key_control" output="screen"/>

    <node pkg="demo06_test_flow_p" type="test01_turtle_spawn_p.py" name="turtle_spawn" output="screen"/>

    <node pkg="demo06_test_flow_p" type="test02_turtle_tf_pub_p.py" name="tf_pub1" args="turtle1" output="screen"/>
    <node pkg="demo06_test_flow_p" type="test02_turtle_tf_pub_p.py" name="tf_pub2" args="turtle2" output="screen"/>
    <node pkg="demo06_test_flow_p" type="test03_turtle_tf_sub_p.py" name="tf_sub" output="screen"/>

</launch>

---

The effect achieved is as follows:

Reference

http://www.autolabor.com.cn/book/ROSTutorials/di-2-zhang-ros-jia-gou-she-ji/23-fu-wu-tong-xin/224-fu-wu-tong-xin-zi-ding-yi-srv-diao-yong-b-python.html