The ompl_lemur package includes the ompl_lemur::Roadmap class, along with several implemented instantiations (e.g. grids, RGGs, Halton graphs, etc). See the Roadmaps page for details.

The or_lemur package includes some tools for working with roadmaps over the C-space of robots in OpenRAVE.

Generating Roadmaps

First, you need to determine the size of your space. or_lemur includes the get-robot-space-bounds.py script:

$ rosrun or_lemur get-robot-space-bounds.py
   --urdf=package://herb_description/robots/herb.urdf
   --srdf=package://herb_description/robots/herb.srdf --manip=right

The output is:

Robot space bounds:
  dof [0] from 0.54159265359 to 5.74159265359 (joint /right/j1, range 5.2)
  dof [1] from -1.96 to 1.96 (joint /right/j2, range 3.92)
  dof [2] from -2.73 to 2.73 (joint /right/j3, range 5.46)
  dof [3] from -0.86 to 3.13 (joint /right/j4, range 3.99)
  dof [4] from -4.79 to 1.3 (joint /right/j5, range 6.09)
  dof [5] from -1.56 to 1.56 (joint /right/j6, range 3.12)
  dof [6] from -2.99 to 2.99 (joint /right/j7, range 5.98)
Bounds command line (e.g. for ompl_lemur generate-roadmap):
  --dim=7 --bounds=0:0.54159265359,5.74159265359 --bounds=1:-1.96,1.96 --bounds=2:-2.73,2.73 --bounds=3:-0.86,3.13 --bounds=4:-4.79,1.3 --bounds=5:-1.56,1.56 --bounds=6:-2.99,2.99

The generated command line can then be passed to the generate-roadmap command:

$ rosrun ompl_lemur generate-roadmap
   --dim=7 --bounds=0:0.54159265359,5.74159265359 --bounds=1:-1.96,1.96 --bounds=2:-2.73,2.73 --bounds=3:-0.86,3.13 --bounds=4:-4.79,1.3 --bounds=5:-1.56,1.56 --bounds=6:-2.99,2.99
   --roadmap-type=HaltonOffDens
   --roadmap-param=num_per_batch=10000
   --roadmap-param=gamma_factor=1.0
   --roadmap-param=scaling=log_n
   --roadmap-param=seed=0
   --num-batches=1
   --out-file=my-roadmap.graphml --out-format=graphml

The resulting graph file can the be loaded into any program that supports the GraphML format (e.g. NetworkX in Python).

Saving Cached Roadmaps

or_lemur includes a implementation of cached roadmaps via the or_lemur::RoadmapCached class. This class stores cache files in $OPENRAVE_HOME/or_lemur/roadmap-HASH.bin files, where HASH is a function of both the space dimensions/bounds and the roadmap type/parameters.

Both the or_lemur::LEMUR and or_lemur::FamilyPlanner planners will use cached roadmaps if they are requested. For example, to use the cached version of the HaltonOffDens roadmap type, pass the CachedHaltonOffDens type instead.

One way to generate a cache file is via the following helper script:

$ rosrun or_lemur save-roadmap-cache.py
   --urdf=package://herb_description/robots/herb.urdf
   --srdf=package://herb_description/robots/herb.srdf --manip=right
   --roadmap-type=HaltonOffDens
   --roadmap-param=num_per_batch=10000
   --roadmap-param=gamma_factor=0.9
   --roadmap-param=scaling=log_n
   --roadmap-param=seed=0
   --num-batches=4

This will produce output which looks something like this:

[RoadmapCached.h:107 initialize] Could not find cached roadmap file: or_lemur/roadmap-934dda6ac853e0c39f6cf12ad9c941c5.bin
[LEMUR.cpp:1489] Densifying roadmap to batch [0] ...
[LEMUR.cpp:1489] Densifying roadmap to batch [1] ...
[LEMUR.cpp:1489] Densifying roadmap to batch [2] ...
[LEMUR.cpp:1489] Densifying roadmap to batch [3] ...
[planner_lemur.cpp:248 PlanPath] Saving cached roadmap ...
[RoadmapCached.h:261 save_file] Saving file ...

In this example, the generated file is approximately 170MiB for four batches.