Showing papers by "Nancy M. Amato published in 2000"

Enhancing randomized motion planners: exploring with haptic hints

Abstract: We investigate methods for enabling a human operator and an automatic motion planner to cooperatively solve a motion planning query. Our goal is to develop techniques by which the automatic planner can utilize (easily generated) user-input, and determine 'natural' ways to inform the user of the progress made by the motion planner. We show that simple randomized techniques inspired by probabilistic roadmap methods are quite useful for transforming approximate, user-generated paths into collision-free paths, and describe an iterative transformation method which enables one to transform a solution for an easier version of the problem into a solution for the original problem. We also illustrate that simple visualization techniques can provide meaningful representations of the planner's progress in a 6-dimensional C-space. We illustrate the utility of our methods on difficult problems involving complex 3D CAD models.

Distributed reconfiguration of metamorphic robot chains

Abstract: The problem we address is the distributed reconfiguration of a metamorphic robotic system composed of any number of two dimensional hexagonal modules from specific initial to specific goal configurations. We present a distributed algorithm for reconfiguring a straight chain of hexagonal modules at one location to any intersecting straight chain configuration at some other location in the plane. We prove our algorithm is correct, and show that it is either optimal or asymptotically optimal in the number of moves and asymptotically optimal in the time required for parallel reconfiguration. We then consider the distributed reconfiguration of straight chains of modules to a more general class of goal configurations.

A general performance model for parallel sweeps on orthogonal grids for particle transport calculations

Abstract: The key contribution of this paper is the first general model which can be used to predict the running time of transport sweeps on orthogonal grids for any regular mapping of the grid cells to processors. Our model, which accounts for machine dependent parameters such as computation cost and communication latency, can be used to analyze and compare the effects of various spatial decompositions on the running time of the transport sweep. Insight obtained from the model yields two significant contributions to the theory of optimal transport sweeps on orthogonal grids. First, our model provides a theoretical basis which explains why, and under what circumstances, the column decomposition of the current standard KBA algorithm is superior to the 'balanced' decomposition obtained by classic domain decomposition techniques. Second, our model enables us to identify a new decomposition, we call Hybrid, which proves to be almost as good as, and sometimes superior to, the current standard KBA method. Our analysis covers sweeps in two- and three-dimensional spatial domains, and first considers sweeps in only one direction, and then sweeps involving multiple simultaneous directions. We obtain expressions for the completion time and discuss theoretical results.

Ligand Binding with OBPRM and Haptic User Input: Enhancing Automatic Motion Planning with Virtual Touch

Abstract: In this paper, we present a framework for studying ligand binding which is based on techniques recently developed in the robotics motion planning community. We are especially interested in locating binding sites on the protein for a ligand molecule. Our work investigates the performance of a fully automated motion planner, as well improvements obtained when supplementary user input is collected using a haptic device. Our results applying an obstacle-based probabilistic roadmap motion planning algorithm (\obprm) to some known protein-ligand pairs are very encouraging. In particular, we were able to automatically generate configurations close to, and correctly identify, the true binding site in the three protein-ligand complexes we tested. We find that user input helps the planner, and a haptic device helps the user to understand the protein structure by enabling them to feel the forces which are hard to visualize.

Computing the arrangement of curve segments: divide-and-conquer algorithms via sampling

Abstract: We describe two deterministic algorithms for constructing the arrangement determined by a set of (algebraic) curve segments in the plane. They both use a divide-and-conquer approach based on derandomized geometric sampling and achieve the optimal running time O(n log n + k), where n is the number of segments and k is the number of intersections. The first algorithm, a simplified version of one presented in [I], generates a structure of size O(nloglogn + k) and its parallel implementation runs in time O(log 2 n). The second algorithm is better in that the decomposition of the arrangement constructed has optimal size O(n + k) and it has a parallel implementation in the EREW PRAM model that runs in time O(log a/2 n). The improvements in the second algorithm are achieved by means of an approach that adds some degree of globality to the divide-and-conquer approach based on random sampling. The approach extends previous work by Dehne et al.[7], Deng and Zhu [8] and Kiihn [9], that use small separators for planar graphs in the design of randomized geometric algorithms for coarse grained multicomputers. The approach simplifies other previous geometric algorithms [1, 2], and also has the potential of providing efficient deterministic algorithms for the external memory model. 1 Problem and Previous Work We consider a classical problem in computational geometry: computing the arrangement determined by a set of curve segments in the plane. There has been a considerable amount of work on this problem in the computational geometry community, particularly for line segments. Starting with a first efficient algorithm by -----t'T~as A&M University, College Station, TX. E-malh amato@cs.tamu.edu. This work was supported in part by NSF CAREER Award CCR-9624315, NSF grants IIS-9619850, ACI9872126, EIA-9805823, and EIA-9810937, by DOE (ASCI ASAP, Level 2 and 3) grant B347886, and by the Texas Higher Education Coordinating Board grant ARP-036327-017. ?The Johns Hopkins University, Baltimore, MD. E-maih goodrich@cs .jhu.edu. This work was partially supported by U.S. Army Research Office MURI Grant DAAH04-96-1-0013 and by U.S. National Science Foundation Grant CCR-9732300. SMax-Planck-Institut ftir Informatik, Saarbr/icken, Germany. E-malh ramesGmpi-sb.mpg, de Bentley and Ottman [4], optimal output sensitive algorithms algorithms were obtained using a deterministic approach by ChazeUe and Edelsbrunner [5] and using randomized approaches by Clarkson and Shor [6] and by Mumuley [10]. These optimal algorithms perform O(n logn + k) work, where n is the number of segments and k is the number of pairwise intersections. They can be adapted so that they are output sensitive even when multiple intersection points are allowed (a point where many segments intersect is counted only once). On the other hand, unlike its randomized counterparts in [6, 10], the deterministic algorithm in [5] can only handle line segments. An alternative deterministic algorithm by Amato et a/.[1], which follows a divide-andconquer approach based on derandomization of geometric sampling, has the advantage of being parallelizable. However, it can only handle line segments and pairwise intersection points, and the decomposition of the arrangement that it constructs has size O(n log log n + k), as opposed to the optimal O(n+ k). One more variation on the problem is to report all the intersections while using only a linear amount of work space. The solutions in [6] and [1] can be adapted to achieve this. Alternatively, Balaban [3] proposed an elementary deterministic algorithm to achieve this; however, it does not construct the arrangement, it does not seem to parallelize, and it cannot handle multiple intersection points. The algorithm in [1] uses an approach based on random sampling that refines iteratively by using small samples to divide the problem. This divide-and-conquer approach and also the well-known random incremental construction (RIC) approach date from work by Clarkson and Shor [6]. Unfortunately, unlike the RIC approach, divide-and-conquer most often leads to nonoptimal algorithms, at least as far as the most basic analysis can tell, because the dividing step creates spurious boundaries that increase the complexity of the constructed decomposition of the arrangement. In fact, the literature is plagued with running times that are a factor n e or log c n away from optimal. Some techniques have been used to correct this and obtain optimal algorithms: sparse cuttings, pruning and biased sampling. In particular, the algorithm in [1] achieves optimality through

An adaptive framework for 'single shot' motion planning

Abstract: This paper proposes an adaptive framework for single shot motion planning, i.e., planning without preprocessing. This framework can be used in any situation, and in particular, is suitable for crowded environments in which the robot's free C-space has narrow corridors such as maintainability studies in complex 3D CAD models. Our iterative strategy adaptively selects a planner whose strengths match the current situation, and then, online, switches to a different planner when circumstances change. This requires techniques to evaluate the characteristics of the current query, and a set of planners which are characterized so that we can match the query with the best planner for it. Our experimental results in complex 3D CAD environments show that our strategy solves queries that none of the planners could solve on their own.

Localization based on visibility sectors using range sensors

Abstract: Presents a rapid localization method for mobile robots. Localization, i.e., absolute position measurement, is an important issue since odometer errors render it impossible for any robot to precisely follow a specified trajectory, resulting in a growing difference between the actual configuration and the calculated configuration as the robot travels. Periodic localization is required to correct these errors. We propose a localization method using range sensor data which is based on simple geometric properties of the environment. In many common situations, information regarding the environment is provided a priori for path planning. During processing, the method proposed here utilizes this information to partition the workspace into sectors using simple visibility computations, and a small identifying label is computed for each sector. The localizer analyzes range sensor readings (distances) and extracts characteristic points, which are compared with the pre-computed sector labels to localize the robot, first to a sector, and then to a particular configuration within that sector. Advantages of this two step process are that it is computationally very simple, and that it allows precise localization without any landmarks from any configuration in the environment. This localization method also provides opportunities for the global navigation procedure to analyze and select trajectories in terms of their tolerance to localization errors.

Distributed reconfiguration of hexagonal metamorphic robots in two dimensions

Abstract: The problem addressed in the distributed reconfiguration of a metamorphic robotic system composed of any number of two dimensional hexagonal modules from specific initial to specific goal configurations. The initial configuration considered is a straight chain of modules, while the goal configurations considered satisfy a more general admissibility condition. A centralized algorithm is described for determining whether an arbitrary goal configuration is admissible. The main result of the paper is a distributed algorithm for reconfiguring a straight chain into an admissible goal configuration. Different heuristics are proposed to improve the performance of the reconfiguration algorithm and simulation results demonstrate the use of these heuristics.© (2000) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

Optimization techniques for probabilistic roadmaps

Abstract: Recently, a new class of randomized path planning methods, known as Probabilistic Roadmap Methods (PRMs) have shown great potential for solving complicated high-dimensional problems. PRMs use randomization (usually during preprocessing) to construct a graph (a roadmap) of representative paths in the robot's configuration space. Vertices correspond to collision-free configurations of the robot. An edge exists between two vertices if a path between the two corresponding configurations can be found by a local planning method. PRMs solve many high degree of freedom (dof) motion planning problems. Unfortunately, for some problems running times may still be unacceptably large and solutions sub-optimal. We provide speed and quality optimization strategies applicable in cluttered 3-dimensional workspaces. Speed improvements for roadmap construction are accomplished by parallel processing and faster failure detection techniques. Quality improvements for the roadmap constructed are accomplished by new roadmap building methods which result in roadmaps that better represent the connectivity of the free configuration space. Quality is also improved by building roadmaps iteratively using information gained at each iteration to control and drive following iterations. Since there is no single generally accepted way of judging roadmap quality, several measures are considered.

Predicting performance on SMPs. A case study: the SGI Power Challenge

Abstract: We study the issue of performance prediction on the SGI-Power Challenge, a typical SMP. On such a platform, the cost of memory accesses depends on their locality and on contention among processors. By running a carefully designed suite of microbenchmarks, we provide quantitative evidence that memory hierarchy effects impact performance far more substantially than other phenomena related to contention. We also fit three cost functions based on variants of the BSP model, which do not account for the hierarchy, and a newly defined function F expressed in terms of hardware counters, which captures both memory hierarchy and contention effects. We test the accuracy of all the functions on both synthetic and application benchmarks showing that, unlike the other functions, F achieves an excellent level of accuracy in all cases. Although hardware counters are only available at run-time, we give evidence that function F can still be employed as a prediction tool by extrapolating values of the counters from pilot runs on small input sizes.

SmartApps: An Application Centric Approach to High Performance Computing

Abstract: State-of-the-art run-time systems are a poor match to diverse, dynamic distributed applications because they are designed to provide support to a wide variety of applications, without much customization to individual specific requirements. Little or no guiding information flows directly from the application to the run-time system to allow the latter to fully tailor its services to the application. As a result, the performance is disappointing. To address this problem, we propose application-centric computing, or SMART APPLICATIONS. In the executable of smart applications, the compiler embeds most run-time system services, and a performance-optimizing feedback loop that monitors the application's performance and adaptively reconfigures the application and the OS/hardware platform. At run-time, after incorporating the code's input and the system's resources and state, the SmartApp performs a global optimization. This optimization is instance specific and thus much more tractable than a global generic optimization between application, OS and hardware. The resulting code and resource customization should lead to major speedups. In this paper, we first describe the overall architecture of Smartapps and then present the achievements to date: Run-time optimizations, performance modeling, and moderately reconfigurable hardware.

An Integrated Mobile Robot Path (Re)Planner and Localizer for Personal Robots

Abstract: Personal robotics applications require autonomous mobile robot navigation methods that are safe, robust, and inexpensive. Most of the previous techniques proposed do not meet these competing goals. In this paper, we describe a method for navigation in a known indoor environment, such as a home or office, that requires only inexpensive range sensors. Our framework includes a high-level planner which integrates and coordinates path planning and localization modules with the aid of a module for computing regions which are expected, with high probability, to contain the robot at any given time. The localization method is based on simple geometric properties of the environment which are computed during a preprocessing stage. The roadmap-based path planner enables one to select routes, and sub-goals along those routes, that will facilitate localization and other optimization criteria. In addition, our framework enables one to quickly plan new routes, dynamically, based on the current position as computed by intermediate localization operations. We present simulation and hardware experimental results that illustrate the practicality and potential of our approach.

A Motion Planning Approach to Folding: From Paper Craft to Protein Structure Prediction

Abstract: In this paper, we present a framework for studying folding problems from a motion planning perspective In particular, all folding objects are modeled as tree-like multi-link articulated `robots'', where fold positions correspond to joints and areas that cannot fold correspond to links This formulation allows us to apply recent techniques developed in the robotics motion planning community for articulated objects with many degrees of freedom (many links) to folding problems An important benefit of this approach is that it not only allows us to study foldability questions, such as, can one object be folded (or unfolded) into another object, but also enables us to study the dynamic folding process itself The framework proposed here has application in traditional motion planning areas such as automation, teaching through demonstration, animation, and most importantly, presents a different approach to the most profound problem in computational biology: protein struction prediction Indeed, our preliminary experimental results with traditional paper crafts (eg, box folding) and a relatively small protein are quite promising

An interactive generalized motion simulator (GMS) in an object-oriented framework

Abstract: This paper introduces I-GMS, a dynamic simulator that accommodates various systems of rigid bodies, ranging from a single free flying rigid object to complex linkages such as those needed for robotic systems or human body simulation. I-GMS's object-oriented design provides a generic framework for representing various types of rigid body systems, and exploits virtual functions to apply common kinematic and dynamic functionalities to them. Moreover, I-GMS supports interactive simulation so that it easily incorporates user-input for the on-line editing and modification of trajectories. User-interaction is achieved with the PHANToM haptic device which runs as an integrated part of I-GMS. We demonstrate I-GMS's capability to simulate general multi-rigid-body systems through examples that include a free-falling sphere, a robot manipulator, and a human body model with 36 degrees of freedom (dof). We present a simple example of interactive simulation for a 3-dof robot manipulator.

Interactive dynamic simulation using haptic interaction

Abstract: Describes an interactive dynamic simulator for virtual environments which allows user interaction via a haptic interface. The interactive simulation is performed in our testbed dynamic simulator I-GMS (Interactive Generalized Motion Simulator), which has been developed in an object-oriented framework for simulating motions of free bodies and complex linkages such as those needed for robotic systems or human body simulation. User interaction is achieved by performing push and pull operations via the PHANToM haptic device which runs as on integrated part of I-GMS. We demonstrate the user interaction capability of I-GMS through online editing of trajectories for a 6-DOF robot manipulator.

Faster, More Effective Connection for Probabilistic Roadmaps

Abstract: In this paper, we report on our experience attempting to improve the running times of probabilistic roadmap motion planning methods (prms). We show that signi cant speedups can be obtained with relatively little e ort on the part of the developer by employing new connection strategies and more intelligent ways of invoking and utilizing existing o -the-shelf collision detection packages. We outline general techniques for determining when and which of the techniques we have developed are most useful. We also categorize each as being helpful in either a problem speci c or problem independent way. Many techniques presented are of general usefulness. This research supported in part by NSF CAREER Award CCR-9624315 (with REU Supplement), NSF Grants IIS-9619850 (with REU Supplement), EIA-9805823, and EIA-9810937, by the Texas Higher Education Coordinating Board under grant ARP-036327-017, and by the NCSA at the University at Illinois at Urbana-Champaign. Dale is supported in part by a Department of Education GAANN fellowship.

Linear-Time Triangulation of a Simple Polygon Made Easier Via Randomization

Abstract: We describe a randomized algorithm for computing the trapezoidal decomposition of a simple polygon. Its expected running time is linear in the size of the polygon. By a well-known and simple linear time reduction, this implies a linear time algorithm for triangulating a simple polygon. Our algorithm is considerably simpler than Chazelle's (1991) celebrated optimal deterministic algorithm and, hence, positively answers his question of whether a simpler randomized algorithm for the problem exists. The new algorithm can be viewed as a combination of Chazelle's algorithm and of non-optimal randomized algorithms due to Clarkson et al. (1991) and to Seidel (1991), with the essential innovation that sampling is performed on subchains of the initial polygonal chain, rather than on its edges. It is also essential, as in Chazelle's algorithm, to include a bottom-up preprocessing phase previous to the top-down construction phase.

Merging Physical Manipulatives and Digital Interface in Educational Software

Abstract: In this paper we describe how elementary school students used physical manipulatives in conjunction with the digital interface of educational software for geometry. The blending o f physical manipulatives and digital interface may help them to ov ercome the limits of the representation and interaction modalities of the digital interface.

How Does It Fold? Searching for Folding Pathways using A Motion Planning Approach

Abstract: In this paper, we present a framework for studying folding problems from a motion planning perspective. The version of the motion planning problem we consider is that of determining a sequence of motions to transform an extended, or flat, configuration of a foldable object (the start) into a known folded configuration (the goal). Modeling foldable objects as tree-like multi-link objects allows us to apply recent techniques developed in the robotics motion planning community for articulated objects with many degrees of freedom (many links) to folding problems. An important feature of this approach is that it not only allows us to study foldability questions, such as, can one object be folded (or unfolded) into another object, but most importantly, provides us with another tool for investigating the dynamic folding process itself. For example, the folding sequences, or pathways, found might provide insight about how a protein folds in nature. Or, the inability to generate a folding sequence could offer insight regarding the foldability of a paper or box folding problem. The framework proposed here has application to traditional motion planning areas such as automation, teaching through demonstration, and animation, and most interesting to us, presents a novel approach for studying folding pathways for proteins. We hope our approach, which constructs a folding sequence to the known native fold, will prove complementary to other methods and might offer additional insight into the larger question of protein structure prediction.