The bringup-package for the gazebo simulation that is used to generate training and test datasets for gaussian process regression.
This package intends to
- modify the Taurob Gazebo model to publish ground-truth odometry
- loads the default
empty.world
provided by thegazebo_ros
package - predefines trajectories used to generate datasets
- configure a linear controller to enable traversing of the trajectories
- see the
linear_controller
ros-package for more details
- see the
- disable the
robot_pose_ekf
package invoked intaurob_tracker_control.launch
and use groundtruth odometry instead- the ground truth odometry is published using the
p3d_base_controller
(seetaurob_tacker_flipper.xacro
) - the
p3d_tf_broadcaster_node
echoes the groundtruth to the TF tree
- the ground truth odometry is published using the
TODO: add a .rosinstall
file.
sudo apt install -y -q \
ros-$ROS_DISTRO-move-base ros-$ROS_DISTRO-gmapping ros-$ROS_DISTRO-explore-lite
In order to run gmapping
on both laser scans, they're combined using the scan_merger
node.
Install and build the ira_laser_tools
package.
This package also depends on the linear_controller
package, which is provided in the root of the gpr_loc
repository.
All of the functionality listed above is invoked using a single launch file.
roslaunch gpr_loc_bringup bringup.launch
To set the direction of the training trajectory, supply the parameter dir_ccw:=false
(defaults to true
).
roslaunch gpr_loc_bringup bringup.launch linear_ctrl:=false
NOTE It is recommended to disable the controller for these worlds and use teleop instead (see the optional argument).
roslaunch gpr_loc_bringup bringup.launch sar_world:=true [linear_ctrl:=false]
To any invocation of bringup.launch
, add gui:=true
.
A separate launch file exists for SLAM (using either manual navigation or explore_lite
). To start the SLAM simulation scenario, run
roslaunch gpr_loc_bringup bringup_mapping.launch
which will start a simulation in a maze-like world, enabling the robot to navigate using move_base
and create an occupancy grid map using gmapping
.
rosbag record /odom /ground_truth/odom /velodyne_points /tf /tf_static /clock
- the simulation time -
std_msgs/Time
- ground truth odometry -
nav_msgs/Odometry
- pre-computed mechanical odometry -
nav_msgs/Odometry
- velocity control signals -
geometry_msgs/Twist
- raw 3D LiDAR data -
sensor_msgs/PointCloud2