OMPL 1.0.0 released!
Mark Moll
All OMPL users are recommended to upgrade. New ROS packages will be made available shortly. We do not anticipate any MoveIt! compatibility problems. There is currently a known issue with generating python bindings on OS X 10.10 (Yosemite), but the MacPorts port of OMPL gets around that by using bindings that were pre-generated on OS X 10.9.
Below is detailed list of changes:
• Added many new planners:
• Linear Temporal Logical Planner (LTLPlanner): a planner that finds solutions for kinodynamic motion planning problems where the goal is specified by a Linear Temporal Logic (LTL) specification.
• Fast Marching Tree algorithm (FMT*): a new asymptotically optimal algorithm contributed by Marco Pavone's Autonomous Systems Laboratory at Stanford.
• Coupled Forest of Random Engrafting Search Trees (CForest): a meta-planner that runs several instances of asymptotically optimal planners in different threads. When one thread finds a better solution path, the states along the path are passed on to the other threads.
• Anytime Path Shortening: a generic wrapper around one or more geometric motion planners that repeatedly applies shortcutting and hybridization to a set of solution paths. Any number and combination of planners can be specified, each is run in a separate thread.
• LazyPRM / LazyPRMstar: not entirely new, but completely re-implemented.
• RRT* has a new option to periodically prune parts of the tree that are guaranteed not to contain the optimal solution. This idea was introduced in CForest, but it useful independently of the CForest parallelization. Although pruning is almost always useful, it is disabled by default to preserve the original behavior.
• Created consistent behavior across all planners that can optimize paths. Calls to the solve method of RRT*, PRM*, SPARS, SPARStwo, and LBTRRT will terminate when (1) the planner termination condition is true or (2) the optimization objective is satisfied. To make these planners terminate when any solution is found, you can set the cost threshold for the optimization objective to, e.g., OptimizationObjective::infiniteCost(). For most of these planners, asymptotic (approximate) optimality is only guaranteed when using the PathLengthOptimizationObjective class.
• Most control-based planners can now use steering functions. The user simply needs to override ompl::control::StatePropagator::steer() and ompl::control::StatePropagator::canSteer() in a derived class.
• Several improvements to benchmarking functionality:
• Planner Arena (http://plannerarena.org) has been relaunched and can be used to interactively visualize benchmark results.
• ompl_benchmark_statistics.py can now also parse MoveIt! benchmark log files using the flag --moveit.
• The config files for ompl_benchmark and the OMPL.app GUI can now be used to specify control-based problems and optimization objectives.
• Added ompl::tools::PlannerMonitor class, which periodically prints planner progress properties in a separate thread. Useful for developing / debugging new planners.
• Updated Py++ toolchain. If you previously installed Py++ and have trouble generating the OMPL Python bindings, you may need to run "make installpyplusplus" again.
• Minimum Boost version is now 1.48 and minimum CMake version is now 2.8.7.
• Bug fixes.
All OMPL users are recommended to update to the latest version. Point your browser to http://ompl.kavrakilab.org/download.html to download OMPL.
You can follow the development of OMPL on Bitbucket at http://bitbucket.org/ompl or the mirror on Github at http://github.org/ompl.
--
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)
Jonathan Bohren
Cool stuff! Can you give a link to a little more info on the LTL planner?
Mark Moll
Bhatia, A., MR. Maly, L. E. Kavraki, and M. Y. Vardi, "Motion Planning with Complex Goals", Robotics Automation Magazine, IEEE, vol. 18, no. 3, pp. 55-64, sept., 2011.
http://dx.doi.org/10.1109/MRA.2011.942115
http://www.kavrakilab.org/sites/default/files/bhatia-maly-kavraki-vardi-ram2011.pdf
There is a demo available here:
http://ompl.kavrakilab.org/LTLWithTriangulation_8cpp_source.html
There are some examples of automata corresponding to common LTL specifications included. To generate automata for arbitrary LTL formula, we use the Spot model checking software (http://spot.lip6.fr/wiki/). That Spot bindings are not included since they are still very much “research code”.
Mark Moll
Wojtynek, Michael
Is there more information how to include the planner to MoveIt and replace the old one!? Building the or_ompl with openrave_catkin worked fine, but I don´t get the steps with the catkin environment hook.
Kind regards,
Michael Wojtynek
-----Original Message-----
From: moveit...@googlegroups.com on behalf of Mark Moll
Sent: Mon 10/27/2014 4:44 PM
To: moveit...@googlegroups.com
Subject: OMPL 1.0.0 released!
We are happy to announce a *major* new release of OMPL: version 1.0.0! It includes many new planners and other improvements. The CForest work was done by Javier V Gomez as part of the 2014 Google Summer of Code (see also his blog post on http://ompl.kavrakilab.org/blog). Dave Coleman's Google Summer of Code contributions will be included in the next release. Other exciting additions include the addition of a planner that combines high-level task specifications in LTL with kinodynamic planning (contributed by Matt Maly) and the FMT* planner (contributed by Marco Pavone's group at Stanford).
All OMPL users are recommended to upgrade. New ROS packages will be made available shortly. We do not anticipate any MoveIt! compatibility problems. There is currently a known issue with generating python bindings on OS X 10.10 (Yosemite), but the MacPorts port of OMPL gets around that by using bindings that were pre-generated on OS X 10.9.
Below is detailed list of changes:
. Added many new planners:
. Linear Temporal Logical Planner (LTLPlanner): a planner that finds solutions for kinodynamic motion planning problems where the goal is specified by a Linear Temporal Logic (LTL) specification.
. Fast Marching Tree algorithm (FMT*): a new asymptotically optimal algorithm contributed by Marco Pavone's Autonomous Systems Laboratory at Stanford.
. Coupled Forest of Random Engrafting Search Trees (CForest): a meta-planner that runs several instances of asymptotically optimal planners in different threads. When one thread finds a better solution path, the states along the path are passed on to the other threads.
. Anytime Path Shortening: a generic wrapper around one or more geometric motion planners that repeatedly applies shortcutting and hybridization to a set of solution paths. Any number and combination of planners can be specified, each is run in a separate thread.
. LazyPRM / LazyPRMstar: not entirely new, but completely re-implemented.
. RRT* has a new option to periodically prune parts of the tree that are guaranteed not to contain the optimal solution. This idea was introduced in CForest, but it useful independently of the CForest parallelization. Although pruning is almost always useful, it is disabled by default to preserve the original behavior.
. Created consistent behavior across all planners that can optimize paths. Calls to the solve method of RRT*, PRM*, SPARS, SPARStwo, and LBTRRT will terminate when (1) the planner termination condition is true or (2) the optimization objective is satisfied. To make these planners terminate when any solution is found, you can set the cost threshold for the optimization objective to, e.g., OptimizationObjective::infiniteCost(). For most of these planners, asymptotic (approximate) optimality is only guaranteed when using the PathLengthOptimizationObjective class.
. Most control-based planners can now use steering functions. The user simply needs to override ompl::control::StatePropagator::steer() and ompl::control::StatePropagator::canSteer() in a derived class.
. Several improvements to benchmarking functionality:
. Planner Arena (http://plannerarena.org) has been relaunched and can be used to interactively visualize benchmark results.
. ompl_benchmark_statistics.py can now also parse MoveIt! benchmark log files using the flag --moveit.
. The config files for ompl_benchmark and the OMPL.app GUI can now be used to specify control-based problems and optimization objectives.
. Added ompl::tools::PlannerMonitor class, which periodically prints planner progress properties in a separate thread. Useful for developing / debugging new planners.
. Updated Py++ toolchain. If you previously installed Py++ and have trouble generating the OMPL Python bindings, you may need to run "make installpyplusplus" again.
. Minimum Boost version is now 1.48 and minimum CMake version is now 2.8.7.
. Bug fixes.
Mark Moll
The latest Indigo packages of OMPL are for version 1.0.0. MoveIt! doesn’t expose every planner in OMPL. This page describes how you can do this yourself:
http://moveit.ros.org/wiki/OMPL/Add_New_Planner
Just out of curiosity, when you say “the planner”, which planner are you referring to?
Mark Moll
Wojtynek, Michael
Hello Mark,
actually I meant planners instead of only one planner. I´m interested in planners which are considering collision avoidance/octomap in their path planning such as the LazyPRM. I´m going to try them all.
To add the planners I´ve downloaded the git OMPL and openrave packages and I did "catkin_make install" after building the packages. Then I´ve added the planners manually to my ompl_planning.yaml. Seems to work!
Kind regards,
Michael Wojtynek
-----Ursprüngliche Nachricht-----
Von: Mark Moll im Auftrag von Mark Moll
Gesendet: Mo 24.11.2014 19:19
An: Wojtynek, Michael
Cc: moveit...@googlegroups.com
Betreff: Re: OMPL 1.0.0 released!