PA Fiducials slides |

Introduction

• This final programming assignment pulls together what we’ve learned up to now
• You will practice detecting fiducials and using them to do simple navigation
• You will have to make sure the camera sees the fiducial and then navigates to it
• The task seems quite simple but it requires mastery of several aspects of Robotics at once
• What you learn here will very likely be directly applicable to your projects.
• Note that this assignment will only be done IRL.

Expected preparation

• Remember that the Platform and BranBot robots are preconfigured for fiducials where as the turtlebots will need some packages installed.
• Before tackling the full assignment, experiment with the fiducials IRL interactively
• You have to write a test program that prints out information of the fiducial it sees
• You can choose to use OpenCV’s fiducial algorithms by themselves or use the aruco_detect layer
• Re-read PRR Chapter 2 to refresh your memory of coordinates and tfs/

Additional references and resources

• Ubiquity Robotics Fiducial Follow
• Ubiquity Robotics Fiducials
• More generic CV: OpenCV aruco detection
• ROS Oriented: aruco_detect
• ROS Wiki tutorials on the tf2 package.
• This may help you understand transforms better.
• The tf cheatsheet in the labnotebook

Requirements for the PA

• Setup
  • Four red blocks in a square with about 2m edges (you can tweak this if you need to)
  • Fiducial taped to each of the blocks, facing inwards
  • Robot in the center of the square
• roslaunch your package
  • The robot spins in place until it seems one (or more) fiducials
  • It drives slowly up to one, and returns to the center
  • And does it for the other three `

Steps

1. Study and understand the mapper.py file.
2. Change the fid_ids parameter of the mapper.py file to match the ids of your fiducials.
3. Complete the nav_real.py file such that the robot:
   1. rotates in place to scan all the target fiducials; and
   2. moves to each target fiducial in turn.
4. Complete the nav_sim.py file such that the robot:
   1. rotates in place to scan all the target fiducials; and
   2. moves to each target fiducial in turn.

The Skeleton Code (Real)

1. bringup (onboard the robot)
2. roslaunch fiducial_nav fiducials_real.launch (on your vnc)
3. rosrun rviz rviz (on your vnc. Tune the RViz to visualize what you want.)
4. rosrun fiducial_nav nav_real.py

• Sample Solution - IRL

What to submit

• A correctly structured ROS package including
  • Nicely commented python source files
  • Readme how to run it
  • Appropriate launch file
• Video of your program running
  • IRL
  • Best to put the video on youtube and give the link
  • In your own voice, explain how the code works