IEEE Robotics & Automation Magazine - September 2020 - 140

●●

●●

●●

 e will now be able to use RViZ (http://wiki.ros.org/rviz)
W
to visualize the robot model (http://wiki.ros.org/rviz/DisplayTypes/RobotModel), showing the links of the CENTAURO robot as defined in the Unified Robot
Description Format (URDF) (http://wiki.ros.org/urdf)
specified in the centauro_basic.yaml configuration
file in the correct poses according to the tf (http://wiki.ros.
org/tf) transform tree. XBotCore runs the robot_state_publisher (http://wiki.ros.org/robot_state_publisher) component
internally so that ROS nodes are able to read the robot transforms from the tf topic. Additionally, the robot's URDF is
published to the ROS parameter server under the name /
xbotcore/robot_description.
Moreover, thanks to the XBot, we are also able to read the
state of the robot's joint using the custom ROS message
(https://github.com/ADVRHumanoids/xbot
_msgs/blob/master/msg/JointState.msg ),
which is published by the communication handler in the /
xbotcore/joint_state.
Once XBotCore is started and the robot is correctly
visualized on RViZ, we are able to start one of the RT
plug-ins listed in the centauro_basic.yaml configuration file. For example, we can move the robot to a
"homing" configuration, calling the available "switch"
ROS service ready to use, because of the following
framework:
rosservice call/xbotcore/-
HomingExample_switch 1

●●

 he robot should now be visualized in the homing configT
uration on RViZ. The user can decide to stop the RT plugins and start the control of the robot using a non-RT
framework by doing the following:
rosservice call/xbotcore/-
HomingExample_switch 0
rosservice call/xbotcore/-
XBotCommunicationPlugin_switch 1

●●

T
 he XBotCommunicationPlugin is a special XBot
plug-in that enables the control of the robot through a nonRT master framework specified in the configuration file we
set. In this basic example, we rely on ROS and define the
following custom ROS message to send reference commands to the robot:
https://github.com/ADVRHumanoids/
xbot_msgs/blob/master/msg/
JointCommand.msg

●●

140

*

 n easy-to-run example for this feature employs either 1) a
A
simple ROS node (implemented in C++ or Python) to move
the CENTAURO robot in the joint space using the custom
joint command message previously described or 2) an even

IEEE ROBOTICS & AUTOMATION MAGAZINE

*

SEPTEMBER 2020

simpler command-line publishing of data to the /
xbotcore/command topic:
rostopic pub/xbotcore/command
xbot_msgs/JointCommand "header:
seq: 0
stamp: {secs: 0, nsecs: 0}
frame_id: ''
name: ['torso_yaw']
position: [0.5]
velocity: 0
effort: 0
stiffness: 0
damping: 0
ctrl_mode: [1]
aux_name: ''
aux: 0"

During the 2018 IEEE/Robotics Society of Japan International Conference on Intelligent Robots and Systems, the
full-day tutorial "A Hands-on Tutorial on XBotCore: A
Real-Time Cross-Robot and Cross-Framework Software
Architecture," about the XBot framework was held (https://
www.iros2018.org/tutorials) (https://xbotcoretutorial
.weebly.com/). Readers can find all of the r-elated material
online at https://github.com/ADVRHumanoids/-
tutorial_iros2018.
One of the core functionalities of the XBot framework is its
cross-robot compatibility due to the XBotInterface component,
which dynamically generates the robot API and a kinematic/
dynamic model of it, taking as an input only the URDF, SRDF
(Semantic Robot Description Format) (http://wiki.ros.org/srdf),
and joint-ID map specified in the XBot configuration file. Moreover, because of the R-HAL component, we are able to go from
the kinematic simulation mode (the dummy mode described
previously) to the dynamic simulation mode, for example, using
the Gazebo simulated environment simply by adding the following plug-in in the input URDF:




The aforementioned homing plug-in can easily be run
not only in dummy mode or on the real robots but also in
the Gazebo simulator, as shown in Figure 6. At the same
time, we are able to go from the simulation mode (either
kinematic or dynamic) to the real robot by providing the
configuration file the low-level R-HAL implementation to
load. As an example, the COMAN+ humanoid robot is
capable of executing the same homing plug-in of the
simulated "dummy" CENTAURO (Figure 7) just by
changing the configuration file, which will now load the
COMAN+ URDF, SRDF, joint-ID map, and R-HAL EtherCAT implementation:
IEEE Robotics & Automation Magazine - September 2020

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - September 2020
