Visualizing Worlds

This tutorial explains you how to visualize a world. There are two recommended ways of doing it. One light weight way through RVIZ2 and a more heavy weight way through simulation with multiverse. Let’s load a world first to get started.

import logging
import os

from semantic_world.adapters.urdf import URDFParser 
from semantic_world.utils import get_semantic_world_directory_root

logging.disable(logging.CRITICAL)
apartment = os.path.join(get_semantic_world_directory_root(os.getcwd()), "resources", "urdf", "apartment.urdf")
world = URDFParser.from_file(apartment).parse()

For the RVIZ2 way, ROS2 is needed. A caveat of this approach is that you have to manage the lifecycle of a ROS2 node yourself. We recommend to put the spinning into sperate threads and just shutdown the thread when exiting the system.

from semantic_world.adapters.viz_marker import VizMarkerPublisher
import threading
import rclpy
rclpy.init()

node = rclpy.create_node("semantic_world")
thread = threading.Thread(target=rclpy.spin, args=(node,), daemon=True)
thread.start()

viz = VizMarkerPublisher(world=world, node=node)

When you want to stop visualizing, you have to stop the visualizer and afterwards clean up ROS2.

node.destroy_node()
rclpy.shutdown()

The multiverse way relies on multiverse and is WIP. Do it faster giang.

If you have followed the guide until here, you have probably noticed that we have used the RayTracer to visualize the world a few times. This is a convenient way of visualizing a world inside a notebook, like in these guides, but it is not recommended for normal usage.