This repository holds Chris Dellin's experimental motion planner LEMUR: Lazily Evaluated Marginal Utility Roadmaps. This code was written at the Personal Robotics Lab at the Robotics Institute at Carnegie Mellon University, and was released on Apr 9, 2015 under a BSD license. It is currently under active development. Previous versions of this planner were known as the multiset planner, E8, or family planner.

The code is split into several ROS packages:

Generic implementations of graph algorithms for the Boost Graph Library (BGL). Includes implemetations of lazy search (LazySP), incremental search (Lifelong Planning A*, incremental bidirectional Dijkstra's), partition functions over graphs, and various helpers for BGL and Boost property maps.

Dependencies: boost

Core planner implementation for the Open Motion Planning Library (OMPL). Uses BGL and pr_bgl for the implementation. Includes a Roadmap class, with implemetations for lattices, random geometric graphs (RGGs), Halton graphs, and densified versions thereof.

Dependencies: ompl, pr_bgl

Bindings for the OpenRAVE planning environment. Related to or_ompl. Includes an OpenRAVE module for managing and relating free C-space subsets induced by collision checking.

Dependencies: ompl, openrave, ompl_lemur

Planning bindings for the PrPy python library.

Dependencies: prpy, or_lemur

Various test scripts for the above packages.

Dependencies: ompl, openrave, ompl_lemur, or_lemur