This project contains a set of ROS packages which provide a navigation suite, including path planning on a given map, controlling of the robot to follow this path and obstacle avoidance while driving.
The packages are build with the goal to get a modular, easily extensible framework, where new modules (e.g. kinematic model of a new robot, new obstacle avoidance algorithms, etc) can be added with minimal effort.
The following video demonstrates various supported robot kinematics:
First, install the dependencies. For example, from your workspace root directory:
cd src
git clone https://github.com/cogsys-tuebingen/cslibs_path_planning
git clone https://github.com/cogsys-tuebingen/cslibs_navigation_utilities
cd ..
rosdep install --from-paths -i -r -y src
Then you can either use catkin_make
, catkin_make_isolated
or catkin build to build the software.
This is only a really quick "quick start" guide. For more detailed explanations, see below.
To start the complete navigation project as well as nodes for SLAM and obstacle detection (using 2d laser), simply run
roslaunch navigation_launch navigation.launch
Now everything is set up and your node can connect to the navigate_to_goal action server and send goals.
If you want to set goal poses manually using Rviz (e.g. for testing or demonstration purposes), run
roslaunch navigation_launch rviz_controlled.launch
By default a controller for car-like robots is used (ackermann_purepursuit). To change this, simply set the environment variable ROBOT_CONTROLLER
with the name of the contproller you want to use.
For example, to use the omni-drive orthexp controller:
export ROBOT_CONTROLLER=omni_orthexp
roslaunch navigation_launch navigation.launch
The controller can also be set via the ROS parameter
rosparam set path_follower/controller_type omni_orthexp
Please take a look at the example launch files for how to customize parameters in your launch files.
For more information please refer to the wiki.