Visual object recognition

You can run this tutorial on:

• ROSbot 2.0
• ROSbot 2.0 PRO
• ROSbot 2.0 simulation model (Gazebo)

We have prepared ready to go virtual environment with end effect of following this tutorial. It is available on ROSDS:

Introduction

Objects can be recognized by a robot with use of a vision system. It is based on image characteristics like points, lines, edges colours and their relative positions.

Processing of object recognition consists of two steps. First is teaching and should be executed before main robot operation. During this step object is presented to the vision system, image and extracted set of features are saved as a pattern. Many objects can be presented to the system.

Second step is actual recognition which is executed constantly during robot operation. Every frame of camera is processed, image features are extracted and compared to data set in the memory. If enough features matches the pattern, then the object is recognized.

In this tutorial we will use find_object_2d node from find_object_2d package for both teaching and recognition.

As an image source we will use nodes from astra.launch as in tutorial 1.

Teaching objects

Anything could be an object to recognize, but remember, that the more edges and contrast colours it has, the easier it will be recognized. A piece of paper with something drawn on it would be enough for this tutorial.

First you should run find_object_2d and astra.launch. Node find_object_2d by default subscribe to image topic, you should remap it to topic /camera/rgb/image_raw.

You can use below launch file:

<launch>

    <arg name="rosbot_pro" default="false" />
    <arg name="use_gazebo" default="false" />

    <arg name="teach" default="true"/>
    <arg name="recognize" default="false"/>

    <arg if="$(arg teach)" name="chosen_world" value="rosbot_world_teaching"/>
    <arg if="$(arg recognize)" name="chosen_world" value="rosbot_world_recognition"/>

    <!-- Gazebo -->
    <group if="$(arg use_gazebo)">
        <include file="$(find rosbot_gazebo)/launch/$(arg chosen_world).launch"/>
        <include file="$(find rosbot_description)/launch/rosbot_gazebo.launch"/>
            <param name="use_sim_time" value="true" />
    </group>

    <!-- ROSbot 2.0 -->
    <group unless="$(arg use_gazebo)">
        <include file="$(find rosbot_ekf)/launch/all.launch">
            <arg name="rosbot_pro" value="$(arg rosbot_pro)" />
        </include>

        <include unless="$(arg use_gazebo)" file="$(find astra_launch)/launch/astra.launch"/>
    </group>

    <node unless="$(arg use_gazebo)" pkg="tf" type="static_transform_publisher" name="camera_publisher" args="-0.03 0 0.18 0 0 0 base_link camera_link 100" />

    <node name="teleop_twist_keyboard" pkg="teleop_twist_keyboard" type="teleop_twist_keyboard.py" output="screen"/>

    <node pkg="find_object_2d" type="find_object_2d" name="find_object_2d">
        <remap from="image" to="/camera/rgb/image_raw"/>
        <param name="gui" value="$(arg teach)"/>
        <param if="$(arg recognize)" name="objects_path" value="$(find tutorial_pkg)/image_rec/"/>
    </node>

    <node pkg="tutorial_pkg" type="action_controller_node" name="action_controller" output="screen"/>

</launch>

To start teaching objects on ROSbot:

roslaunch tutorial_pkg tutorial_4.launch use_gazebo:=false teach:=true recognize:=false

To start teaching objects in Gazebo:

roslaunch tutorial_pkg tutorial_4.launch use_gazebo:=true teach:=true recognize:=false

To place objects in front of camera using Gazebo, you can use buttons Translation mode and Rotation mode in top left corner and drag objects to desired position. Another option is to drive ROSbot to look at the selected object.

After launching the find_object_2d node with properly adjusted image topic, new window should appear:

On the left side of the window there are thumbnails of saved images (should be empty at first run). Application main window contains camera view. Yellow circles on the image are marking extracted image features.

To begin teaching process choose from the main toolbar Edit → Add object from scene.... New window will appear:

Now move the object and camera in order to cover as many features of the object as possible. While doing that try not to catch object surroundings. When it’s done, click Take picture.

In next view click Select region and select only that part of taken picture, that covers desired object and click Next.

You will get confirmation of features extracted from selected image area. If presented image is in accordance with what you selected, click End

You should see new thumbnail in the left panel. Notice the number outside of parentheses on the left of the image, this is the object ID.

Now you can add some more objects to be recognized. Remember their IDs, you will need them later:

When you have enough objects in the database choose from the main toolbar File → Save objects... and choose a folder to store recognized objects. Close the window and stop all running nodes.

Recognizing objects

Objects will be recognized by the same node which was used for teaching but it works in slightly different configuration. We will set two new parameters for the node. For parameter gui we will set value false, this will run node without window for learning objects as we no longer need it. Another parameter will be objects_path, this should be a path to a folder that you have just chosen to store recognized objects.

You can use the same launch file as for teaching, but with different parameter values.

On ROSbot:

roslaunch tutorial_pkg tutorial_4.launch use_gazebo:=false teach:=false recognize:=true

In Gazebo:

roslaunch tutorial_pkg tutorial_4.launch use_gazebo:=true teach:=false recognize:=true

Node is publishing to /objects topic with message type std_msgs/Float32MultiArray. Data in this message is covering: object ID, size of recognized object and its orientation. The most interesting for us is object ID, this is first element of array.

Whenever object is recognized, it will be published in that topic. Place different objects in front of camera and observe as their data is published in the topic.

To watch messages published in the topic, you can use rostopic tool:

rostopic echo /objects

Making decision with recognized object

To perform a robot action based on recognized object, we will make a new node. It’s task will be to subscribe /objects topic and publish message to /cmd_vel with speed and direction depending on the object.

Create a new file, name it action_controller.cpp and place it in src folder under tutorial_pkg. Then open it in text editor and paste below code:

#include <ros/ros.h>
#include <std_msgs/Float32MultiArray.h>
#include <geometry_msgs/Twist.h>

#define SMILE 4
#define ARROW_LEFT 3
#define ARROW_UP 5
#define ARROW_DOWN 6

int id = 0;
ros::Publisher action_pub;
geometry_msgs::Twist set_vel;

void objectCallback(const std_msgs::Float32MultiArrayPtr &object)
{
   if (object->data.size() > 0)
   {
      id = object->data[0];

      switch (id)
      {
      case ARROW_LEFT:
         set_vel.linear.x = 0;
         set_vel.angular.z = 1;
         break;
      case ARROW_UP:
         set_vel.linear.x = 1;
         set_vel.angular.z = 0;
         break;
      case ARROW_DOWN:
         set_vel.linear.x = -1;
         set_vel.angular.z = 0;
         break;
      default: // other object
         set_vel.linear.x = 0;
         set_vel.angular.z = 0;
      }
      action_pub.publish(set_vel);
   }
   else
   {
      // No object detected
      set_vel.linear.x = 0;
      set_vel.angular.z = 0;
      action_pub.publish(set_vel);
   }
}

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

   ros::init(argc, argv, "action_controller");
   ros::NodeHandle n("~");
   ros::Rate loop_rate(50);
   ros::Subscriber sub = n.subscribe("/objects", 1, objectCallback);
   action_pub = n.advertise<geometry_msgs::Twist>("/cmd_vel", 1);
   set_vel.linear.x = 0;
   set_vel.linear.y = 0;
   set_vel.linear.z = 0;
   set_vel.angular.x = 0;
   set_vel.angular.y = 0;
   set_vel.angular.z = 0;
   while (ros::ok())
   {
      ros::spinOnce();
      loop_rate.sleep();
   }
}

Below is an explanation of the code line by line.

Including required headers:

#include <ros/ros.h>
#include <std_msgs/Float32MultiArray.h>
#include <geometry_msgs/Twist.h>

Defining constants for recognized objects, adjusting values to IDs of objects recognized by your system:

#define SMILE 8
#define ARROW_LEFT 9
#define ARROW_UP 10
#define ARROW_DOWN 11

Integer for storing object identifier:

int id = 0;

Publisher for velocity commands:

ros::Publisher action_pub;

Velocity command message:

geometry_msgs::Twist set_vel;

Callback function for handling incoming messages with recognized objects data:

void objectCallback(const std_msgs::Float32MultiArrayPtr &object) {

Checking if size of data field is non zero, if it is, then object is recognized. When data field size is zero, then no object was recognized.

if (object->data.size() > 0) {

Reading id of recognized object:

id = object->data[0];

Depending on recognized object, setting appropriate speed values:

switch (id)
{
case ARROW_LEFT:
   set_vel.linear.x = 0;
   set_vel.angular.z = 1;
   break;
case ARROW_UP:
   set_vel.linear.x = 1;
   set_vel.angular.z = 0;
   break;
case ARROW_DOWN:
   set_vel.linear.x = -1;
   set_vel.angular.z = 0;
   break;
default: // other object
   set_vel.linear.x = 0;
   set_vel.angular.z = 0;
}

Publishing velocity command message:

action_pub.publish(set_vel);

Stopping all motors when no object was detected:

else
{
   // No object detected
   set_vel.linear.x = 0;
   set_vel.angular.z = 0;
   action_pub.publish(set_vel);
}

Main function, node initialization and setting main loop interval:

int main(int argc, char **argv)
{
   ros::init(argc, argv, "action_controller");
   ros::NodeHandle n("~");
   ros::Rate loop_rate(50);

Subscribing to /objects topic:

ros::Subscriber sub = n.subscribe("/objects", 1, objectCallback);

Preparing publisher for velocity commands:

action_pub = n.advertise<geometry_msgs::Twist>("/cmd_vel", 1);

Setting zeros for initial speed values:

set_vel.linear.x = 0;
set_vel.linear.y = 0;
set_vel.linear.z = 0;
set_vel.angular.x = 0;
set_vel.angular.y = 0;
set_vel.angular.z = 0;

Main loop, waiting for trigger messages:

while (ros::ok())
{
   ros::spinOnce();
   loop_rate.sleep();
}

Last thing to do is editting the CMakeLists.txt file. Find line:

add_executable(tutorial_pkg_node src/tutorial_pkg_node.cpp)

and add below code after it:

add_executable(action_controller_node src/action_controller.cpp)

Find also:

target_link_libraries(tutorial_pkg_node
    ${catkin_LIBRARIES}
)

and add below code after it:

target_link_libraries(action_controller_node
    ${catkin_LIBRARIES}
)

Now you can build your node and test it.

Task 1 Run your node along with find_object_2d and astra.launch or Gazebo simulator. Then use rosnode, rostopic and rqt_graph tools to examine the system. Place different objects in front of your robot one by one. Observe how it drives and turns depending on differnt objects.

You can use below launch file:

<launch>

    <arg name="rosbot_pro" default="false" />
    <arg name="use_gazebo" default="false" />

    <arg name="teach" default="false"/>
    <arg name="recognize" default="true"/>

    <arg if="$(arg teach)" name="chosen_world" value="rosbot_world_teaching"/>
    <arg if="$(arg recognize)" name="chosen_world" value="rosbot_world_recognition"/>

    <!-- Gazebo -->
    <group if="$(arg use_gazebo)">
        <include file="$(find rosbot_gazebo)/launch/$(arg chosen_world).launch"/>
        <include file="$(find rosbot_description)/launch/rosbot_gazebo.launch"/>
            <param name="use_sim_time" value="true" />
    </group>

    <!-- ROSbot 2.0 -->
    <group unless="$(arg use_gazebo)">
        <include file="$(find rosbot_ekf)/launch/all.launch">
            <arg name="rosbot_pro" value="$(arg rosbot_pro)" />
        </include>

        <include unless="$(arg use_gazebo)" file="$(find astra_launch)/launch/astra.launch"/>
    </group>

    <node unless="$(arg use_gazebo)" pkg="tf" type="static_transform_publisher" name="camera_publisher" args="-0.03 0 0.18 0 0 0 base_link camera_link 100" />

    <node name="teleop_twist_keyboard" pkg="teleop_twist_keyboard" type="teleop_twist_keyboard.py" output="screen"/>

    <node pkg="find_object_2d" type="find_object_2d" name="find_object_2d">
        <remap from="image" to="/camera/rgb/image_raw"/>
        <param name="gui" value="$(arg teach)"/>
        <param if="$(arg recognize)" name="objects_path" value="$(find tutorial_pkg)/image_rec/"/>
    </node>

    <node pkg="tutorial_pkg" type="action_controller_node" name="action_controller" output="screen"/>

</launch>

Getting position of recognized object

Besides the ID of recognized object, find_object_2d node is also publishing a homography matrix of recognized object. In computer vision homography is used to define position of object relative to the camera. We will use it to obtain horizontal position of the object. Homography matrix is published in the same topic as the ID, but in the next cells of array, they are formatted as

[objectId1, objectWidth, objectHeight, h11, h12, h13, h21, h22, h23, h31, h32, h33, object2].

We will modify node to rotate robot to the direction of recognized object.

Open action_controller.cpp file in text editor.

Begin with including of required header file:

#include <opencv2/opencv.hpp>

Variable for storing camera centre- this should be half of your camera horizontal resolution:

int camera_center = 320; // left 0, right 640

Variables for defining rotation speed:

float max_ang_vel = 0.6;
float min_ang_vel = 0.4;
float ang_vel = 0;

Variable for object width and height:

float objectWidth = object->data[1];
float objectHeight = object->data[2];

Variable for storing calculated object centre:

float x_pos;

Variable defining how much rotation speed should increase with every pixel of object displacement:

float speed_coefficient = (float) camera_center / max_ang_vel /4;

Object for homography matrix:

cv::Mat cvHomography(3, 3, CV_32F);

Vectors for storing input and output planes:

std::vector<cv::Point2f> inPts, outPts;

Adding new case in switch statement:

case SMILE:

Extracting coefficients homography matrix:

cvHomography.at<float>(0, 0) = object->data[3];
cvHomography.at<float>(1, 0) = object->data[4];
cvHomography.at<float>(2, 0) = object->data[5];
cvHomography.at<float>(0, 1) = object->data[6];
cvHomography.at<float>(1, 1) = object->data[7];
cvHomography.at<float>(2, 1) = object->data[8];
cvHomography.at<float>(0, 2) = object->data[9];
cvHomography.at<float>(1, 2) = object->data[10];
cvHomography.at<float>(2, 2) = object->data[11];

Defining corners of input plane:

inPts.push_back(cv::Point2f(0, 0));
inPts.push_back(cv::Point2f(objectWidth, 0));
inPts.push_back(cv::Point2f(0, objectHeight));
inPts.push_back(cv::Point2f(objectWidth, objectHeight));

Calculating perspective transformation:

cv::perspectiveTransform(inPts, outPts, cvHomography);

Calculating centre of object from its corners:

x_pos = (int) (outPts.at(0).x + outPts.at(1).x + outPts.at(2).x + outPts.at(3).x) / 4;

Calculating angular speed value proportional to position of object and putting it into velocity message:

ang_vel = -(x_pos - camera_center) / speed_coefficient;

if (ang_vel >= -(min_ang_vel / 2) && ang_vel <= (min_ang_vel / 2))
{
   set_vel.angular.z = 0;
}
else if (ang_vel >= max_ang_vel)
{
   set_vel.angular.z = max_ang_vel;
}
else if (ang_vel <= -max_ang_vel)
{
   set_vel.angular.z = -max_ang_vel;
}
else
{
   set_vel.angular.z = ang_vel;
}

Your final file should look like this:

#include <ros/ros.h>
#include <std_msgs/Float32MultiArray.h>
#include <geometry_msgs/Twist.h>
#include <std_msgs/String.h>
#include <opencv2/opencv.hpp>

#define SMILE 4
#define ARROW_LEFT 3
#define ARROW_UP 5
#define ARROW_DOWN 6
int id = 0;
ros::Publisher action_pub;
geometry_msgs::Twist set_vel;
int camera_center = 320; // left 0, right 640
float max_ang_vel = 0.6;
float min_ang_vel = 0.4;
float ang_vel = 0;

void objectCallback(const std_msgs::Float32MultiArrayPtr &object)
{
   if (object->data.size() > 0)
   {
      id = object->data[0];

      float objectWidth = object->data[1];
      float objectHeight = object->data[2];
      float x_pos;
      float speed_coefficient = (float)camera_center / max_ang_vel / 4;

      // Find corners OpenCV
      cv::Mat cvHomography(3, 3, CV_32F);
      std::vector<cv::Point2f> inPts, outPts;
      switch (id)
      {
      case ARROW_LEFT:
         set_vel.linear.x = 0;
         set_vel.angular.z = 1.0;
         break;
      case ARROW_UP:
         set_vel.linear.x = 1;
         set_vel.angular.z = 0;
         break;
      case ARROW_DOWN:
         set_vel.linear.x = -1;
         set_vel.angular.z = 0;
         break;
      case SMILE:
         cvHomography.at<float>(0, 0) = object->data[3];
         cvHomography.at<float>(1, 0) = object->data[4];
         cvHomography.at<float>(2, 0) = object->data[5];
         cvHomography.at<float>(0, 1) = object->data[6];
         cvHomography.at<float>(1, 1) = object->data[7];
         cvHomography.at<float>(2, 1) = object->data[8];
         cvHomography.at<float>(0, 2) = object->data[9];
         cvHomography.at<float>(1, 2) = object->data[10];
         cvHomography.at<float>(2, 2) = object->data[11];

         inPts.push_back(cv::Point2f(0, 0));
         inPts.push_back(cv::Point2f(objectWidth, 0));
         inPts.push_back(cv::Point2f(0, objectHeight));
         inPts.push_back(cv::Point2f(objectWidth, objectHeight));
         cv::perspectiveTransform(inPts, outPts, cvHomography);

         x_pos = (int)(outPts.at(0).x + outPts.at(1).x + outPts.at(2).x +
                       outPts.at(3).x) /
                 4;
         ang_vel = -(x_pos - camera_center) / speed_coefficient;

         if (ang_vel >= -(min_ang_vel / 2) && ang_vel <= (min_ang_vel / 2))
         {
            set_vel.angular.z = 0;
         }
         else if (ang_vel >= max_ang_vel)
         {
            set_vel.angular.z = max_ang_vel;
         }
         else if (ang_vel <= -max_ang_vel)
         {
            set_vel.angular.z = -max_ang_vel;
         }
         else
         {
            set_vel.angular.z = ang_vel;
         }

         break;
      default: // other object
         set_vel.linear.x = 0;
         set_vel.angular.z = 0;
      }
      action_pub.publish(set_vel);
   }
   else
   {
      // No object detected
      set_vel.linear.x = 0;
      set_vel.angular.z = 0;
      action_pub.publish(set_vel);
   }
}

int main(int argc, char **argv)
{
   std_msgs::String s;
   std::string str;
   str.clear();
   str.append("");
   std::to_string(3);
   s.data = str;
   ros::init(argc, argv, "action_controller");
   ros::NodeHandle n("~");
   ros::Subscriber sub = n.subscribe("/objects", 1, objectCallback);
   ros::Rate loop_rate(50);
   action_pub = n.advertise<geometry_msgs::Twist>("/cmd_vel", 1);
   set_vel.linear.x = 0;
   set_vel.linear.y = 0;
   set_vel.linear.z = 0;
   set_vel.angular.x = 0;
   set_vel.angular.y = 0;
   set_vel.angular.z = 0;
   while (ros::ok())
   {
      ros::spinOnce();
      loop_rate.sleep();
   }
}

Last thing to do is to edit the CMakeLists.txt file. Find line:

find_package(catkin REQUIRED COMPONENTS roscpp )

and add below code after it:

find_package( OpenCV REQUIRED )

Find also:

include_directories(
    ${catkin_INCLUDE_DIRS}
)

and change it to:

include_directories(
    ${catkin_INCLUDE_DIRS}
    ${OpenCV_INCLUDE_DIRS}
)

Find:

target_link_libraries(action_controller_node
    ${catkin_LIBRARIES}
)

and change it to:

target_link_libraries(action_controller_node
    ${catkin_LIBRARIES}
    ${OpenCV_LIBRARIES}
)

Now you can build your node and test it.

Task 2 Run your node along with find_object_2d and astra.launch or Gazebo simulator. Place the object with ID bonded to new case in switch statement in front of your robot. Observe how it turns towards the object.

Following the object

Open action_controller.cpp file in text editor.

Begin with including required header file:

#include <sensor_msgs/Range.h>

Add variables for measured object distance, average distance and desired distance to obstacle:

float distFL = 0;
float distFR = 0;
float average_dist = 0;
float desired_dist = 0.2;

Callback functions for incoming sensor messages, their task is only to put values into appropriate variables:

void distFL_callback(const sensor_msgs::Range &range)
{
   distFL = range.range;
}

void distFR_callback(const sensor_msgs::Range &range)
{
   distFR = range.range;
}

Then, in switch statement, calculate average distance and set velocity proportional to it only if both sensors found an obstacle, else set zero value for linear velocity:

if (distFL > 0 && distFR > 0)
{
   average_dist = (distFL + distFR) / 2;
   set_vel.linear.x = (average_dist - desired_dist) / 4;
}
else
{
   set_vel.linear.x = 0;
}

In main function, subscribe to sensor topics:

ros::Subscriber distFL_sub = n.subscribe("/range/fl", 1, distFL_callback);
ros::Subscriber distFR_sub = n.subscribe("/range/fr", 1, distFR_callback);

Final file should look like this:

#include <ros/ros.h>
#include <std_msgs/Float32MultiArray.h>
#include <std_msgs/Int32MultiArray.h>
#include <geometry_msgs/Twist.h>
#include <std_msgs/String.h>
#include <opencv2/opencv.hpp>
#include <sensor_msgs/Range.h>

#define SMILE 4
#define ARROW_LEFT 3
#define ARROW_UP 5
#define ARROW_DOWN 6
int id = 0;
ros::Publisher action_pub;
geometry_msgs::Twist set_vel;
int camera_center = 320; // left 0, right 640
float max_ang_vel = 0.6;
float min_ang_vel = 0.4;
float ang_vel = 0;

float distFL = 0;
float distFR = 0;
float average_dist = 0;
float desired_dist = 0.2;

void distFL_callback(const sensor_msgs::Range &range)
{
   distFL = range.range;
}

void distFR_callback(const sensor_msgs::Range &range)
{
   distFR = range.range;
}

void objectCallback(const std_msgs::Float32MultiArrayPtr &object)
{
   if (object->data.size() > 0)
   {
      id = object->data[0];

      float objectWidth = object->data[1];
      float objectHeight = object->data[2];
      float x_pos;
      float speed_coefficient = (float)camera_center / max_ang_vel;

      // Find corners OpenCV
      cv::Mat cvHomography(3, 3, CV_32F);
      std::vector<cv::Point2f> inPts, outPts;
      switch (id)
      {
      case ARROW_LEFT:
         set_vel.linear.x = 0;
         set_vel.angular.z = 1;
         break;
      case ARROW_UP:
         set_vel.linear.x = 1;
         set_vel.angular.z = 0;
         break;
      case ARROW_DOWN:
         set_vel.linear.x = -1;
         set_vel.angular.z = 0;
         break;
      case SMILE:
         cvHomography.at<float>(0, 0) = object->data[3];
         cvHomography.at<float>(1, 0) = object->data[4];
         cvHomography.at<float>(2, 0) = object->data[5];
         cvHomography.at<float>(0, 1) = object->data[6];
         cvHomography.at<float>(1, 1) = object->data[7];
         cvHomography.at<float>(2, 1) = object->data[8];
         cvHomography.at<float>(0, 2) = object->data[9];
         cvHomography.at<float>(1, 2) = object->data[10];
         cvHomography.at<float>(2, 2) = object->data[11];

         inPts.push_back(cv::Point2f(0, 0));
         inPts.push_back(cv::Point2f(objectWidth, 0));
         inPts.push_back(cv::Point2f(0, objectHeight));
         inPts.push_back(cv::Point2f(objectWidth, objectHeight));
         cv::perspectiveTransform(inPts, outPts, cvHomography);

         x_pos = (int)(outPts.at(0).x + outPts.at(1).x + outPts.at(2).x +
                       outPts.at(3).x) /
                 4;
         ang_vel = -(x_pos - camera_center) / speed_coefficient;

         if (ang_vel >= -(min_ang_vel / 2) && ang_vel <= (min_ang_vel / 2))
         {
            set_vel.angular.z = 0;
            if (distFL > 0 && distFR > 0)
            {
               average_dist = (distFL + distFR) / 2;
               set_vel.linear.x = (average_dist - desired_dist) / 4;
            }
            else
            {
               set_vel.linear.x = 0;
            }
         }
         else if (ang_vel >= max_ang_vel)
         {
            set_vel.angular.z = max_ang_vel;
         }
         else if (ang_vel <= -max_ang_vel)
         {
            set_vel.angular.z = -max_ang_vel;
         }
         else
         {
            set_vel.angular.z = ang_vel;
         }

         break;
      default: // other object
         set_vel.linear.x = 0;
         set_vel.angular.z = 0;
      }
      action_pub.publish(set_vel);
   }
   else
   {
      // No object detected
      set_vel.linear.x = 0;
      set_vel.angular.z = 0;
      action_pub.publish(set_vel);
   }
}

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

   std_msgs::String s;
   std::string str;
   str.clear();
   str.append("");
   std::to_string(3);
   s.data = str;
   ros::init(argc, argv, "action_controller");
   ros::NodeHandle n("~");
   ros::Subscriber sub = n.subscribe("/objects", 1, objectCallback);

   ros::Subscriber distL_sub = n.subscribe("/range/fl", 1, distFL_callback);
   ros::Subscriber distR_sub = n.subscribe("/range/fr", 1, distFR_callback);
   ros::Rate loop_rate(50);
   action_pub = n.advertise<geometry_msgs::Twist>("/cmd_vel", 1);
   set_vel.linear.x = 0;
   set_vel.linear.y = 0;
   set_vel.linear.z = 0;
   set_vel.angular.x = 0;
   set_vel.angular.y = 0;
   set_vel.angular.z = 0;
   while (ros::ok())
   {
      ros::spinOnce();
      loop_rate.sleep();
   }
}

Now you can build your node and test it.

Task 3 Run your node along with find_object_2d and astra.launch or Gazebo simulator. Place the same object as in Task 2 in front of your robot. Observe how it turns and drives towards the object.

Summary

After completing this tutorial you should be able to configure your CORE2 device with vision system to recognize objects. You should also be able to determine the position of recognized object relative to camera and create a node that perform specific action related to recognized objects. You should also know how to handle proximity sensors with sensor_msgs/Range message type.

---

by Łukasz Mitka, Husarion

Do you need any support with completing this tutorial or have any difficulties with software or hardware? Feel free to describe your thoughts on our community forum: https://community.husarion.com/ or to contact with our support: support@husarion.com