/movo_object_search

Primary LanguagePython

NOTE:

THiS REPOSITORY IS A COPY FROM zkytony/kinova-movo. FOR REFERENCE ONLY. FOR DEVELOPMENT, BUILD UPON THE KINOVA-MOVO REPO.

Package for MOVO Object Search

Creating this package

catkin_create_pkg movo_object_search ar_track_alvar std_msgs rospy roscpp pcl_ros pcl_msgs
catkin_make -DCATKIN_WHITELIST_PACKAGES="movo_object_search"

Start up movo

roslaunch movo_bringup movo_system.launch

Run movo RVIZ

roslaunch movo_viz view_robot.launch

Note that to visualize the point cloud, go to Sensing-> PointCloud2, and make sure the Topic is `/movo_camera/point_cloud/points

How-tos

How to start a Gazebo simulation

  1. roslaunch movo_object_search world_bringup.launch

    This command will start the Gazebo simulation (by default, the gui is true), and other necessary MOVO components. When the Gazebo simulation starts, wait for a few seconds until the MOVO dance completes.

    By default the environment is worlds/tabletop_cube.sdf which is located under the movo_gazebo package. If you want to change it, edit the default value of the parameter world in environment.launch.

How to start navigation stack

  1. roslaunch movo_object_search navigation.launch

    This will launch the necessary components for the navigation stack; It is just a proxy for $(find movo_demos)/launch/nav/map_nav.launch file.

    By default the map for the tabletop_cube environment is used; The name of the map is called test_simple, and the .pgm and .yaml files are located under $(find movo_demos)/map/. It's very confusing that the world models and the map files are not in the same package. But, it is what it is.

Notes:

  • The amcl poses are published under /amcl_pose, not /robot_pose.
  • The map is served at /map topic, as usual.
  • In the topological map package, even though there is a camera pose topic argument, the code actually does not make use of the camera pose. Therefore I commented it out in the launch file for topological map generation.

How to get volumetric observation?

  1. Go to $(movo_object_search)/scripts/observation and run python process_pcl.py. If you want to detect AR tag as well python process_pcl.py -M. Note that due to ROS-related reasons, python process_pcl.py -M will only broadcast observations if an AR tag is detected. Therefore, I recommend running these two together separately, each with a different node name.

Which RVIZ to use?

  1. Use the nav_cloud.rviz which allows navigation and also visualizes the observation and point cloud.