Grasping is one of the more advanced (and probably harder to use)
features of OpenRAVE. The tutorial and examples will hopefully shed a
lot of light into what functionality is available and how easy it is
to use. This is a hot area of research, and the tutorial by no means
solves all the problems associate with finding good grasps for
objects. But it is a start.
In other recent updates,
OpenRAVE now has a human model with very highly articulated fingers
(thanks to James Kuffner). You can find it at robots/man1.robot.xml.
The hands are in robots/man1rhand.robot.xml. Soon, there will be a
demo script (similar to the Hanoi problem) showing how to plan for two
robots autonomously playing chess while using the automatically
generated grasp tables (unlike the Hanoi problem).
There have been also some changes to how a parameters are passed to a
planner in order to allow the scripts to directly call a planner. The
parameters can now be serialized to XML (and vice-versa). Combined
with the logging plugin, it is now possible to save a snapshot of a
plan and the scene to be rerun at any point in time without going
through all the code used to generate the state.