/ros_control_boilerplate

Provides a simple simulation interface and template for setting up a hardware interface for ros_control

Primary LanguageC++BSD 3-Clause "New" or "Revised" LicenseBSD-3-Clause

ROS Control Boilerplate

Simple simulation interface and template for setting up a hardware interface for ros_control. The idea is you take this as a starting point for creating your hardware interfaces, and it is needed because ros_control documentation is sparse. This boilerplate demonstrates:

  • Creating a hardware_interface for multiple joints for use with ros_control
  • Position Trajectory Controller
  • Control of 2 joints of the simple robot "RRBot" pictured below
  • Loading configurations with roslaunch and yaml files
  • Generating a random trajectory and sending it over an actionlib interface
  • Partial support of joint mode switching (needs to be improved)
  • Joint limits
  • Pass-through non-physics based robot simulator
  • Visualization in Rviz

This open source project was developed at PickNik Robotics. Need professional ROS development and consulting? Contact us at projects@picknik.ai for a free consultation.

Maintainers

Special thanks to the following maintainers of this repo:

  • Dave Coleman (@davetcoleman)
  • Andy Zelenak (@AndyZe)
  • John Morris (@zultron)
  • Robert Wilbrandt (@RobertWilbrandt)

Status:

  • Build Status Travis CI
  • Devel Job Status Devel Job Status
  • Build Status AMD64 Debian Job Status

Video Demo

See YouTube for a very modest video demo.

Install

This package depends on gazebo_ros_demos for its rrbot_description package, but you must add it to your catkin workspace by source:

git clone https://github.com/ros-simulation/gazebo_ros_demos.git

Then, either install this package from source so you can develop off of it, or install from debian:

sudo apt-get install ros-indigo-ros-control-boilerplate

Run Simulation Demo

This package is setup to run the "RRBot" two joint revolute-revolute robot demo. This "template package" is located in the ros_control_boilerplate as a subfolder that you can easily rename and reuse. To run its ros_control non-physics-based simulated hardware interface, run:

roslaunch ros_control_boilerplate rrbot_simulation.launch

To visualize its published /tf coordinate transforms in Rviz run:

roslaunch ros_control_boilerplate rrbot_visualize.launch

To send a random, dummy trajectory to execute, run:

roslaunch ros_control_boilerplate rrbot_test_trajectory.launch

Customize

To test this as a simulation interface for your robot, you can quickly rename the subfolder package into the name of your robot using the following commands:

function findreplace() {
    grep -lr -e "$1" * | xargs sed -i "s/$1/$2/g" ;
}

function findreplacefilename() {
    find . -depth -name "*$1*" -exec bash -c 'for f; do base=${f##*/}; mv -- "$f" "${f%/*}/${base//'$1'/'$2'}"; done' _ {} +
}

findreplacefilename rrbot myrobot
findreplace rrbot myrobot
findreplace RRBot MyRobot
findreplace RRBOT MYROBOT

Then add the necessary code to communicate with your robot via USB/serial/ethernet/etc in the file myrobot_hw_interface.cpp.

Setting an Initial Position, Using with MoveIt!

If you need your robot to startup at a particular position in simulation, or you would like to use this funcitonality to simulate your robot with MoveIt!, see the downstream package (it depends on this package) moveit_sim_controller

Other Helper Tools

Recording to CSV

Write the commands from a trajectory controller to csv file

rosrun ros_control_boilerplate controller_to_csv SAVE_TO_FILE_PATH CONTROLLER_STATE_TOPIC TIME_TO_RECORD

Commanding from CSV

Read from csv file and execute on robot

rosrun ros_control_boilerplate csv_to_controller READ_FROM_FILE_PATH CONTROLLER_STATE_TOPIC TIME_TO_RECORD

Commanding from Keyboard

Joint-level teleop from a keyboard (TODO: remove had coded topic names)

rosrun ros_control_boilerplate keyboard_teleop

Limitations

  • Does not implement estops, transmissions, or other fancy new features of ros_control
  • Does not have any hard realtime code, this depends largely on your platform, kernel, OS, etc
  • Only position control is fully implemented, though some code is in place for velocity control

Contribute

Please add features, make corrections, and address the limitations above, thanks!