Introduction

The goals of this lab are to gain experience working with PID controllers, launch files, and the ROS parameter server.

Turtlebot PID

• Log into one of the Turtlebot notebooks. Follow the steps from the ROS Packages Lab to create a repository named pid_chase on stu. Check the repository out to your local machine. Don't create a package at this point.
• Unzip the file pid_chase.zip and copy the contents into your pid_chase repository. If you've done this correctly, your pid_chase package should now be configured as follows:

  $ cd ~/cs354_ws/src/pid_chase/
  $ ls
  CMakeLists.txt  launch  package.xml  scripts  src
  

• Take a minute to read over the file pid_chase.py in the scripts directory, and the file pid_chase.launch in the launch directory. Make sure you understand the contents of those files.
• Build your workspace and call source devel/setup.bash.
• Connect a Turtlebot to your notebook. For this lab you will need to connect both the Kinect and the robot itself. The Kinect should be attached to the USB port on the right side of the computer. Launch the Turtlebot using pid_chase.launch. Point the robot toward a wall and run pid_chase.py. You should see an error like the following:

  [ERROR] [WallTime: 1422370481.748347] PID Parameters not set.  Exiting
  

• Use the rosparam command to set the three parameters expected by pid_chase.py to some initial values and re-execute pid_chase.py.

• Experiment with different values for your gain parameters until you find settings that work well. You want the Turtlebot to smoothly approach the target distance then stop at the appropriate point without excessive oscillations. The p_error term should be driven to a value that is very close to 0. The pid_chase.py node publishes all three error terms. You can echo them to a terminal:

  rostopic echo /p_error
  

  or you can visualize them using rqt_plot:

  rosrun rqt_plot rqt_plot /p_error/data /i_error/data
  

• Once you are satisfied with your gain parameters, modify pid_chase.launch to set the parameters appropriately. The launch file XML format is described on this wiki page. Test the new launch file. Once everything is working, commit your changes and push your work to stu.
  

If You Have Extra Time

• One weakness of the current version of this code is that the robot makes fast velocity changes when the distance changes: for example, when someone steps in front of the depth sensor. This results in jerky motions that are hard on the robot's motors.

  We can force the velocity to change smoothly by taking a weighted average between the previous velocity and the desired velocity:

        new_velocity = α × desired_velocity + (1 - α) × previous_velocity

  Values of α near 1 will lead to fast velocity changes, while values of α near 0 will lead to slow changes in velocity.

• Update pid_chase.py to implement velocity smoothing. (You may need to re-tune your PID gain terms. In effect, smoothing the velocity changes the dynamics of the system that the PID controller is trying to control.)