LEMUR Planner

Here is top-level documentation for how to use the implementation of LEMUR for OMPL.

The primary planner is ompl_lemur::LEMUR, a derived class of ompl::base::Planner. It takes the following arguments:

• const ompl::base::SpaceInformationPtr & si
• ompl_lemur::TagCache & tag_cache

The roadmap to be searched is an instance of a roadmap type which subclasses ompl_lemur::Roadmap. Some of the classes implemented are described on the Roadmaps page!

For an overview, see the main page.

Parameters

In addition to the construction arguments above, LEMUR takes the following parameters:

Roadmap

• roadmap_type (see LEMUR::registerRoadmapType)
• roadmap.<roadmap_param>: delegated to roadmap instance

Optimization coefficients

• coeff_distance (double)
• coeff_checkcost (double)
• coeff_batch (double)

These parameters specify the weights on the objective function optimized by the planner at each iteration.

Search type: search_type (string)

This parameter specifies which algorithm is used for the inner search performed by LazySP. Currently supported values include:

• dijkstras (rooted at start)
• astar (rooted at start)
• lpastar (rooted at start)
• incbi

Evaluation selector type: eval_type (string)

This parameter specifies which edge selector is used to select edges for evaluation at each iteration. Currently supported values include:

• fwd
• rev
• alt
• bisect
• fwd_eval
• partition_all
• sp_indicator_probability

Other parameters

• do_timing (bool): Set to true to have the planner profile and report time spent during search and during edge evaluation.
• persist_roots (bool): Set to true to keep vertices and edges from previous problem definitions
• num_batches_init (int): number of batches to generate before proceeding with search
• max_batches (int): planner terminates after search fails over this number of batches generated