[ompl-users] OMPL 0.10.0 released

[Mark M. <mm...@ri...>]

We are happy to announce a new release of OMPL, version 0.10.0. All OMPL users are recommended to update to the new version. It includes the following changes:

• Minimum Boost version is now 1.44.
• Windows support
• Added a primer on sampling-based motion planning and OMPL.
• Added an implementation of SyCLoP, a new multilayered meta-planner that combines discrete search with a sampling-based motion planner. Implementations of Syclop with RRT and EST as the low-level planners are provided.
• Added an implementation of EST for planning with controls.
• New generic interface for parameter settings for almost anything that is user-configurable. This was done to simplify reading parameters from an input file (see next bullet).
• New benchmark program that allows one to easily specify benchmark problems with simple configuration files.
• New state spaces for Dubins vehicles and Reeds-Shepp vehicles. This allows one to use any geometric planner to compute feasible paths for such vehicles. See the demo program demo_GeometricCarPlanning.
• Added state serialization, deserialization, state space signatures.
• Added the notion of DirectedControlSampler.
• Added path hybridization, a technique for combining several solutions to a motion planning query to form a better bath. We also introduced a utility class called ParallelPlan that runs several planner instances in parallel and (optionally) uses path hybridization.
• Added a more advanced path shortcutting method.
• Added support for FCL collision checking library. PQP is actually still faster, and is therefore still the default.
• All versions of EST and SBL use the new PDF class to sample more efficiently from empirical probability distribution functions.
• Added a wrapper for boost::numeric::odeint to support high-order numerical integration of ordinary differential equations.
• Fixed issues with static variables for console.
• Added RPY sampling to the RNG class.
• Made some public members protected for GoalState and GoalStates; improvements to GoalLazySamples.
• Added support for multiple solution paths in a Goal datastructure.
• Added StateSpace::getValueAddressAtLocation() to allow faster indexing in states using state space names.
• The C++ code switched to the preferred boost::function syntax. A general method for exposing boost::function objects to Python was added, but this requires a patch to pygccxml.
• To make installing and patching Py++, pygccml, and gccxml easier there is now a script called installPyPlusPlus.sh that does this. One can call this script after running cmake by typing “make installpyplusplus”.

Point your browser to http://ompl.kavrakilab.org/download.html to download the latest version.

-- 
Mark Moll
http://www.cs.rice.edu/~mmoll
Rice University, MS 132, PO Box 1892, Houston, TX 77251
713-348-5934 (phone), 713-348-5930 (fax)