ICRA 2014 Roundup

Mark Moll

A couple weeks ago the IEEE International Conference on Robotics and Automation took place in Hong Kong. There were many papers that used OMPL one way or another. Here is an overview of some of them:

  1. Asymptotically near-optimal RRT for fast, high-quality, motion planning
    Oren Salzman and Dan Halperin
    Blavatnik School of Computer Science, Tel-Aviv University, Israel
    This paper describes Lower Bound Tree-RRT (LBT-RRT), a single-query sampling-based algorithm that is asymptotically near-optimal. By varying the approximation factor one can get behavior that ranges from regular RRT to RRT*. An implementation of this algorithm has been contributed by the authors.

  2. Fast Stochastic Motion Planning with Optimality Guarantees using Local Policy Reconfiguration
    Ryan Luna, Morteza Lahijanian, Mark Moll, and Lydia E. Kavraki
    Department of Computer Science, Rice University
    This paper presents a framework for fast reconfiguration of local control policies for a stochastic system to satisfy a high-level task specification. It uses Bounded Markov Decision Processes at the high-level. Some follow-up work will be presented at AAAI 2014 and at WAFR 2014. See http://www.kavrakilab.org/biblio.

  3. Improving Efficiency of Intricate Manipulation Planning through Mapping of Grasp Feasibility Zones
    Mihai Pomârlan [1] and Ioan A. Sucan [2]
    1 Universitatea Politehnica Timisoara, facultatea ETC
    2 Willow Garage (now at Google[X])
    This paper is on pre-computing (off-line) a ranking of grasping poses for each configuration in a roadmap constructed for the object to be manipulated. This allows choosing more likely grasping poses when manipulating known objects and leads to faster computation of manipulation plans. The paper uses the LazyPRM and RRTConnect implementations from OMPL.

  4. AUV Mission Control via Temporal Planning
    Michael Cashmore [1], Maria Fox [1], Tom Larkworthy [2], Derek Long [1], and Daniele Magazzeni [1]
    1 King’s College London, UK
    2 Heriot-Watt University, Edinburgh, UK
    This paper combines high-level temporal planning with low-level motion planning. As the title suggests, they apply this to AUVs and have some nice experimental results.

  5. SMT-Based Synthesis of Integrated Task and Motion Plans from Plan Outlines
    Srinivas Nedunuri, Sailesh Prabhu, Mark Moll, Swarat Chaudhuri and Lydia E. Kavraki
    Department of Computer Science, Rice University
    In this paper the authors propose an integrated task and motion planning framework called ROBOSYNTH that translates high-level task specifications in the form of plan outlines to detailed low-level plans. In a plan outline there are “holes” that ROBOSYNTH will fill in. It can, e.g., determine a low-level path for picking up an object subject to user constraints or a feasible order in which dishes could be loaded in the dishwasher. This is done by combining an SMT solver with OMPL.

  6. A Sampling-Based Strategy Planner for Nondeterministic Hybrid Systems
    Morteza Lahijanian, Lydia E. Kavraki, and Moshe Y. Vardi
    Department of Computer Science, Rice University
    This paper presents a strategy planner for hybrid systems with nondeterministic discrete transitions. Examples of such systems include a car-like robot with faulty transmission (gearbox) causing nondeterministic switching between gears.

  7. Motion Planning for Robotic Manipulators with Independent Wrist Joints
    Kalin Gochev [1], Venkatraman Narayanan [2], Benjamin Cohen [1], Alla Safonova [1] Maxim Likhachev [2]
    1 Department of Computer and Information Science, University of Pennsylvania
    2 Robotics Institute, Carnegie Mellon University
    This paper proposes to decompose high-dimensional planning problems into lower-dimensional subproblems and solve them with deterministic heuristic search methods. The approach is compared to OMPL and is shown to have some advantages to some planner in OMPL on some metrics. It discusses trade-offs in path quality, speed, and completeness.

  8. Gaussian Process Kernels for Rotations and 6D Rigid Body Motions
    Muriel Lang, Oliver Dunkley and Sandra Hirche
    Technische Universität München
    This paper extends Gaussian Processes to quaternions and dual quaternions, which is useful for learning the dynamics of rotational or rigid body motion, respectively. The approach is shown to be superior to using Euler angles. OMPL is used to generate trajectories on which the approach is tested.

  9. Interactive-rate Motion Planning for Concentric Tube Robots
    Luis G. Torres, Cenk Baykal, and Ron Alterovitz
    Department of Computer Science, University of North Carolina
    The authors are interested in planning minimally invasive surgery using concentric tube robots. The approach combines a precomputed roadmap with an iterative IK solver to obtain interactive-rate planning times.

  10. Poisson-RRT
    Chonhyon Park and Jia Pan and Dinesh Manocha
    Department of Computer Science, University of North Carolina
    [project page]
    The authors present an RRT-based motion planning algorithm that uses the maximal Poisson-disk sampling scheme. The approach exploits the free-disk property of the maximal Poisson-disk samples to generate nodes and perform tree expansion. The authors demonstrate that the approach can be parallelized on CPUs as well as GPUs.

If you use OMPL in your work, please let us know! We are always interested to how and where it is used.