World canvas server

A component of the world canvas framework, a ROS stack for storing and accessing semantic information about the world, with an initial emphasis on needs and use cases for mobile robots.

World canvas server is a storage manager for semantic maps. Initial version replicates map_store behavior for semantic maps.

Give it a fast try

Installing

Use world_canvas.rosinstall to create a world_canvas workspace. Don't forget to install dependencies:

rosdep install --from-paths -i src

Running

First of all, launch the annotations server:

roslaunch world_canvas_server world_canvas_server.launch debug:=true --screen

Then you need to populate the database with some annotations (we still don't have an editor :( Use the save_xxxx.py scripts and the test/annotations/xxx_list.yaml data files from world_canvas_server package, e.g.

> rosrun world_canvas_server save_markers.py _world:='What a wonderful world' _file:=`rospack find world_canvas_server`/test/annotations/ar_list.yaml
> rosrun world_canvas_server save_walls.py _world:='What a wonderful world' _file:=`rospack find world_canvas_server`/test/annotations/wall_list.yaml

These scripts will request the server to save in database both the annotations and the associated data the YAML files contain.

And now you are ready to look for annotations. Before giving a try to our client libraries, you can use the get_any.py script from world_canvas_server package, e.g.

> rosrun world_canvas_server get_any.py  _world:='What a wonderful world' _ids:=[] _types:=['ar_track_alvar_msgs/AlvarMarker'] _keywords:=[] _relationships=[] _topic_type:=ar_track_alvar_msgs/AlvarMarker _topic_name:=ar_markers _pub_as_list:=False
> rosrun world_canvas_server get_any.py  _world:='What a wonderful world' _ids:=[] _types:=['yocs_msgs/Wall'] _keywords:=[] _relationships=[] _topic_type:=yocs_msgs/WallList _topic_name:=wall_pose_list _pub_as_list:=True

The first 5 parameters provide search criteria:

• world Retrieved annotations associated to this world. Mandatory
• ids Retrieved annotation uuid is within this list
• types Retrieved annotation type is within this list
• keywords Retrieved annotation has AT LEAST one of these keywords
• relationships Still not implemented on server, so just ignored

And the other 3 define how the server must publish the retrieved annotations

• topic_name Where server must publish annotations data
• topic_type The message type to publish annotations data. Mandatory if pub_as_list is true; ignored otherwise
• pub_as_list If true, annotations will be packed in a list before publishing, so topic_type must be an array of the requested annotations

The get_any script calls 3 services in the annotations server:

• get annotations to retrieve the annotations that satisfy the filter criteria
• get annotations data to retrieve the data of those annotations
• pub annotations data to make the server publish such data

You can verify that all went right by echoing topics

• topic_name Published by the server with the loaded annotations data
• topic_name + '_client' Published by the get_Any script with the retrieved annotations data

Or in RViz showing topic

• topic_name + '_markers' Published by the get_Any script with visualization markers

Import/export database

The save_xxxx.py scripts are a temporal workaround to play with world canvas. The canonical ways to populate the database are a graphic tool still do be implemented and an import from YAML files service. To try the later method, and the reverse operation of exporting to file, you can run the import.py/export.py scripts:

rosrun world_canvas_server import.py _file:=`rospack find world_canvas_server`/test/annotations/full_db.yaml
rosrun world_canvas_server export.py _file:=`rospack find world_canvas_server`/test/annotations/export_db.yaml

Real scenario testing

Now we can show a more practical use: Turtlebot simulated on Stage using virtual sensor and getting the 2D map from WC server. You must first update your workspace using the modified rosinstall file. It adds the experimental branch of the turtlebot_simulator repo. Then, populate the WC database, and ready to go!

roslaunch world_canvas_server world_canvas_server.launch --screen
roslaunch turtlebot_stage populate_world.launch --screen
roslaunch turtlebot_stage world_canvas_demo.launch --screen