Home

Awesome

or_urdf

Build Status

or_urdf is an OpenRAVE plugin for loading a URDF and, optionally, SRDF file as an OpenRAVE KinBody or Robot. This package provides the OpenRAVE URDF module through a plugin. This plugin can be instantiated using the following Python code:

module = RaveCreateModule(env, 'urdf')

Once the module has been instantiated, you can use the module to load a KinBody by calling the custom load command with a path to a URDF file. Similarly, you can create an OpenRAVE Robot calling the load command with paths to both URDF and SRDF files.

Why URDF and SRDF?

URDF and SRDF are the standard XML file formats used to describe robots in ROS. These files are available for many robots and are a key requirement of using MoveIt! and/or RViz. This module uses the standard urdfdom and srdfdom parsers and, thus, should be compatible with any robot specification that obeys the URDF and SRDF standards.

Why programmatic construction?

or_urdf takes the unique approach of progammatically constructing OpenRAVE objects from the URDF and SRDF files. This is in contrast to other conversion functions, like the urdf_to_collada script, which converts the input URDF into an intermediate file format. Programmatically constructing the OpenRAVE objects has several key advantages over this alternative approach:

  1. Relative file:// and package:// URI's are resolved at run-time.
  2. There is no need to re-generate any auto-generated files when the URDF or SRDF files change.
  3. There is no loss in precision due to the serialization and deserialization of floating point numbers.
  4. The URDF and SRDF specifications can be loaded directly from the robot_description and semantic_robot_description ROS parameters.

Loading a KinBody from URDF

The following code will load the URDF model /path/to/my/model.urdf as an OpenRAVE KinBody:

with env:
    name = module.SendCommand('load /path/to/my/model.urdf')
    body = env.GetKinBody(name)

The following OpenRAVE properties have no equivalent in URDF and, thus, must be manually configured on a KinBody created by or_urdf:

The load command programmatically creates a KinBody from the URDF by building a list of LinkInfo and JointInfo structures. This KinBody is then added to the environment with the anonymous flag set. The command returns the name of the object that was added to the environment. This is necessary to avoid introducing name conflicts or ambiguity when loading multiple instances of a URDF file into the same environment.

You can easily change the name of the resultant KinBody if the name generated by this automatic procedure are undesirable. To do so, you must: (1) remove the KinBody from the environment, (2) change the KinBody's name, and (3) add the KinBody back to the environment. For example:

with env:
    name = module.SendCommand('load /path/to/my/model.urdf')
    body = env.GetKinBody(name)
    
    env.Remove(body)
    body.SetName('my_custom_name')
    env.Add(body)

Loading a Robot from URDF and SRDF

or_urdf also supports loading an OpenRAVE Robot from the combination of a URDF and SRDF file. In this case, the URDF file fills the role of an OpenRAVE .kinbody.xml file and the SRDF file fills the role of an OpenRAVE .robot.xml file. Unfortunately, there is not a direct mapping between the SRDF file format and the features supported by OpenRAVE. or_urdf performs the conversion as follows:

The following OpenRAVE properties have no equivalent in SRDF and, thus, must be manually configured on a Robot create by or_urdf:

Just as with creating a KinBody, or_urdf programmatically creates the Robot by constructing LinkInfo, JointInfo, and ManipulatorInfo structs. The following code creates an OpenRAVE Robot from the paired /path/to/my/model.urdf and /path/to/my/model.srdf files:

with env:
    name = module.SendCommand('load /path/to/my/model.urdf /path/to/my/model.srdf')
    body = env.GetRobot(name)

See above for more information about how to rename the robot created by the module.

License

or_urdf is licensed under a BSD license. See LICENSE for more information.

Contributors

or_urdf was developed by the Personal Robotics Lab in the Robotics Institute at Carnegie Mellon University. This library is developed and maintained by Michael Koval and Pras Velagapudi.