Using Floor Fiducials To Guide The Robot

We are adding a powerful ability to our fiducial follow application that will allow users to write a python or other program to send fiducial follow as well as basic movement commands to the application through a ROS topic. We hope to officially release this code in the first quarter of 2020.

Be aware that this node is under development and as recently as the beginning of February we have decided to convert this node to use the standard ‘actionlib’ mechanism used widely in ROS. So as of early February our goal is to have a different ROS node that will be started with the same launch file we discuss in this document. The sample client will change to be an actionlib client.

This page describe usage of a launch file that once released places the robot into a mode where it can be issued commands to find and approach or move about in simple movements. Contact us on the forum if you require this ability prior to introduction and we can discuss Beta testing this ability. We call this launch file fiducial_floor_follow.launch and it assumes fiducials of size 110mm because this size seems a good choice for floor based fiducials given camera field of view and so on.

A Few Words To Help Visualize How This Mode Enables User Applications

The enhanced app is written in Python. The existing code for the fiducial_follow app may be found on THIS_PAGE .

Several commands can be used in applications where floor-attached fiducials are an acceptable solution. The commands allow functions such as the following:

With these commands some sequential factory robotic applications can be implemented without full implementation of the much more complex full room navigation methods utilizing our MoveBasic command modes. In this way a robot can follow a trail of fiducials on the floor based on a user application sending commands as markers are reached.

This provides a simpler mode that may in many cases be all that is required.

The ROS topic of /follower_commands is issued commands with a name and parameters depending on the action. ROS topic /follower_status will offer feedback to the user program. So a program can say to drive to a fiducial then wait for positive status before issuing another command. In this way a user application can have it’s own logic and let the Magni be the ‘server’ to perform the movement operations. A user program once at a given location could for example then tell it’s own custom hardware to do some sort of load or unload operation using the custom user hardware that may have been designed.

We offer a simple python script that can talk to and listen to this new node and we call that script which is very basic but perhaps a good head start to develop your own specific control required to suit your end application.

Setting Up a Downward Facing Camera to Use This App

By default we will specify that a downward facing RaspiCam will be used and will be mounted 55cm above the floor and tilted 45 degrees downward ahead of the robot (these numbers may change). We also are thinking floor fiducials should be 110mm across for many reasons but if required users can modify this as suit their needs.

We plan on making camera mounting hardware to properly release this mode. A user will have to edit /etc/ubiquity/robot.yaml as root and change position to ‘downward’:

     position: 'downward'

The Git Code Branch for the Fiducial Floor Follow App

This mode of fiducial follow as of the start of 2020 is not yet released but the code will be under our ‘demos’ git repository in a git branch called ‘floor_follow’. More files like the launch file and changed URDF file are in our magni_robot repository in the branch also called ‘floor_follow’. If desired users who are familiar with the git revision control system can contact us and we can offer direction on how to use this mode as a Beta Site until it becomes released in first quarter of 2020 (our goal).

Running the Fiducial Floor Follow App

The Ubiquity Robotics robots use fiducial markers (aruco markers). The fiducial_floor_follow launch for the app takes advantage of aruco_detect (a ROS node that is part of the Magni package) to detect aruco markers in the image feed from a camera. The characteristics of the markers’ images enable the robot to compute its location and that of the markers.

To launch this application the following line will be used in a console

roslaunch magni_demos fiducial_floor_follow.launch

While this launch file is running the console will show that aruco_detect continuously searches its field of view for aruco markers. When a marker is found and reported to the app the app knows or rather ‘sees’ the fiducial. You can use RVIZ to see what the camera sees and adjust lighting if required so that you can verify that the fiducial is really being seen by messages on the console.

Documents that describe in the level of detail required to develop or use this new mode are in our ‘demos/fiducial_follow’ repository in branch ‘floor_follow’. Refer to THIS LOCATION.

This is an exciting new mode that will offer the ability for users to develop custom applications more easily. We have an example python client that shows how to drive this code over the ROS topic and it is called found in above repository and git branch.