Open Motion Planning Library

business

License: New and Simplified BSD licenses

Web Page: http://kavrakilab.org/GSoC2014.html

Mailing List: https://lists.sourceforge.net/lists/listinfo/ompl-users

We are a team of researchers and developers working on efficient implementations of state-of-the-art motion planning algorithms and their integration with other open source software. The library is aimed at robotics researchers, robotics students & educators, and industrial end users. In 2012 OMPL was awarded the Grand Prize in Open Source Software Word Challenge (http://ossaward.org), an open source competition organized by the South Korean Ministry of Knowledge and Economy.

Projects

• Implementation of C-Forest, Bidirectional FMT and their combination My proposal consists on implementing two different modules for the OMPL: a parallel framework for simple-query sampling-based path planners and a new planner, the Bi-directional Fast Marching Tree. In the third and last step of the project, I plan to combine the two previous part into a single optimal planner which will, most probably, outperform the current approaches significantly.
• Improving OMPL Performance with Threading and Experienced-Based Planning Several aspects of OMPL will be improved during this project, as time allows. A multi-threaded RRT* algorithm will be implemented that achieves superlinear speedup for shortest path problems. Cost-based planning will be improved by updating the T-RRT algorithm with recent advances. Experienced-based planning will be added to OMPL to reduce planning time using various cutting edge methods including sampling from constraint manifolds.