E1AR0002@SMUVM1.BITNET (03/24/86)
@device(dover) @libraryfile(SpecialCharacters) The following technical reports are available from Information Services The Robotics Institute Carnegie-Mellon University Pittsburgh, PA 15213. or Serviou@H.CS.CMU.EDU @report[title="Experiences with SRL: An Analysis of a Frame-based Knowledge Representation", author="Mark S. Fox, J. Mark Wright, and David Adam", number="CMU-RI-TR-85-10", date="July 1985"] The goal of this paper is to examine a single representation language, SRL, and its applications to determine utility of its ideas. Post mortems have been performed before but have the appearance of a massive "weeding" due to the plethora of ideas included in the initial version of the language. What distinguishes SRL is its @i<evolution> from a research engine to a "production level" language. Its evolution has been hastened by its application to "real" problems, and its transition to industrial use. @i<(14 pages)> @report[title="The Use of Multiple Problem Decompositions in Time Constrained Planning Tasks", author="Stephen F. Smith, and Peng Si Ow", number="CMU-RI-TR-85-11", date="July 1985"] Problems requiring the synthesis of a collection of plans accomplishing distinct (but mostly related) goals has received increasing attention within AI. Such problems are typically formulated as multi-agent planning problems, emphasizing a problem decomposition wherein individual agents assume responsibility for the generation of individual plans while taking into account the goals and beliefs of other agents in the system. One consequence of such a problem decomposition is a simplified view of resource allocation that assumes avoidance of conflicts to be the sole concern. The validity of this assumption comes into question in time constrained problem domains requiring the allocation of multiple, shared resources. In job shop scheduling, for example, where sequences of manufacturing operations must be determined and scheduled for multiple orders, it is necessary to consider much more than availability to efficiently allocate resources over time. We argue that in such domains, an ability to reason from both resource-based and agent-based perspectives is essential to appropriate consideration of all domain constraints. @i<(6 pages)> @report[title="Planning Robot Grasping Motions in the Presence of Uncertainty", author="Randy C. Brost", number="CMU-RI-TR-85-12", date="July 1985"] This paper presents a method for automatic planning of robot grasping motions that are guaranteed to succeed, despite bounded variations in the object's location. The method capitalizes on the physics of friction to generate the space of all guaranteed grasp plans that utilize either a squeeze-grasp, offset-grasp, or push-grasp motion. All plans that are found are guaranteed to succeed, even if the worst-case error occurs. From this space of guaranteed plans, a plan can be chosen and executed without sensing or feedback. Executing the motion removes two degrees of uncertainty from the object's position. For simplicity, planar motion of the object during grasping is assumed. @i<(71 pages)> @report[title="Optimal Algorithms for Finding the Symmetries of a Planar Point Set", author="P. T. Highnam", number="CMU-RI-TR-85-13", date="August 1985"] We present an asymptotically optimal algorithm to locate all the axes of mirror symmetry of a planar point set. The algorithm was derived by reducing the 2-D symmetry problem to linear pattern-matching. Optimal algorithms for finding rotational symmetries and deciding whether a point symmetry exists are also presented. @i<(6 pages)> @report[title="Optimization of Mechanisms for Force Generation by Using Pulleys and Spring", author="Tokuji Okada", number="CMU-RI-TR-85-14", date="September 1985"] A mechanism is described for achieving a desired force/motion relationship. The mechanism employs two hinged arms with pulleys and springs. In comparison to active force-control methods, the device is compact, energy saving, and robust. The device is ideally suited to miniature devices, and in a recent application, has been used in a mobile robot for inspecting pipes. The relationship between the motion of the mechanism and its output force is analyzed using both analytical and approximate techniques to determine the optimum configuration and the dimensions of the various components. In the final design, experimental results demonstrate the superiority of non-circular eccentric pulleys over conventional pulleys for producing a specified force/motion curve. @i<(31 pages)> @report[title="Accurate Trajectory Control of Robotic Manipulators", author="J. C. Van Winssen and C. W. de Silva", number="CMU-RI-TR-85-15", date="April 1985"] This report presents a control scheme for accurate trajectory following with robotic manipulators. The method uses feedforward control using model-based torques for fast operation and gross compensation, and adaptive feedback control for correcting deviations from the desired trajectory under feedforward control. The adaptive controller eliminates trajectory-following errors in the least squares sense. The control scheme takes into account dynamic nonlinearities (e.g., coriolis and centrifugal accelerations and payload changes), geometric nonlinearities (e.g., nonlinear coordinate-transformation matrices) and physical nonlinearities (e.g., nonlinear damping) as well as dynamic coupling in manipulators. Computer simulations are presented to indicate the effectiveness and robustness of the control scheme. When the desired trajectory is completely known before the control scheme is implemented, then off-line computations can be used to generate the adaptive feedback gains and the computational efficiency will not be a major limiting factor with this control scheme. If real-time changes in the desired trajectory have to be accommodated, the computational efficiency has to be improved using recursive relations to compute the adaptive gains. The necessary recursive relations are derived and presented in this report. @i<(50 pages)> @report[title="DP - Format of the Drawing Files", author="Dario Giuse", number="CMU-RI-TR-85-16", date="September 1985"] DP is a highly interactive graphics editor that runs on a personal workstation and can produce general-purpose illustrations as well as circuit drawings. The main purpose of this document, which is a complete specification of the format and semantics of DP drawing files, is to encourage the development of application programs that can read and generate drawings in the DP format and thus exchange information in graphics form. @i<(23 pages)> @report[title="Representation of Activity Knowledge for Project Management", author="Arvind Sathi, Mark S. Fox, and Mike Greenberg", number="CMU-RI-TR-85-17", date="September 1985"] Representation of activity knowledge is important to any application which must reason about activities, such as new product management, factory scheduling, robot control, vehicle control, software engineering and air traffic control. This article provides an integration of the underlying theories needed for modeling activities. Using the domain of large computer design projects as example, the semantics of activity modeling is described. While past research in Knowledge Representation has discovered most of the underlying concepts, our attempt is toward their integration. This includes the epistemological concepts for erecting the required knowledge structure; the concepts of Activity, State, Goal and Manifestation for the adequate description of the plan and the progress; and the concepts of Time and Causality to infer the progression among the activities. We also address the issues which arise due to the integration of aggregation, time and causality among activities and states. @i<(55 pages)> @report[title="The Motion of a Pushed, Sliding Object Part I: Sliding Friction", author="M. A. Peshkin and A. C. Sanderson", number="CMU-RI-TR-85-18", date="September 1985"] In many robotic applications, manipulation planning for an object free to slide on a surface is an important problem. Physical analysis of the object's motion is made difficult by the absence of information about the distribution of support of the object, and of the resulting frictional forces. This paper describes a new approach to the analysis of sliding motion. The instantaneous motion of the object can be described as a pure rotation about a center of rotation (COR) somewhere in the plane. In this paper we find the locus of CORs for all possible distributions of support forces. We assume zero friction at the pusher-object contact, and we assume that the support force distribution is confined to a disk. In one application to robotic manipulation, bounds on the distance an object must be pushed to come into alignment with a robot finger or a fence are determined. @i<(48 pages)> @report[title="The Use of Quantitative Databases in Aladin, an Alloy Design System", author="Ingemar Hulthage, Michael D. Rychener, Mark S. Fox, and Martha L. Farinacci", number="CMU-RI-TR-85-19", date="September 1985"] Aladin is an expert system that assists metallurgists in the design and evaluation of new alloys. This paper gives a preliminary overview of the architecture of Aladin with emphasis on the coupling of symbolic and numeric computations. We briefly describe the knowledge representation of Alloys, Phase diagrams and Microstructure information. We explain how qualitative reasoning is used to define the variables to be dealt with quantitatively and how regression among known alloys is used. We describe multidimensional constraints and how they are used to deal with interaction between variables to implement a least commitment strategy, and how they at the same time provide criteria for search termination and backtracking. @i<(13 pages)> @report[title="Calibration and Use of a Light Stripe Range Sensor Mounted on the Hand of a Robot", author="Gerald J. Agin", number="CMU-RI-TR-85-20", date="November 1985"] A three-dimensional sensing capability can be very useful in automating the robotic handling of industrial parts. This report describes a sensor that uses a Puma robot to carry a cage in which are mounted a light-stripe projector and a lightweight television camera in fixed relationship to each other. Calculation of the Cartesian coordinates of all points illuminated by the stripe is a straightforward matter, provided the positions of the manipulator end point, the camera, and the projector are accurately known. Calibration is necessary to determine these positions. Since the robot itself introduces significant systematic errors, we present a procedure for reducing these errors by recalibration of the zero reference points for each joint. The relative positions of the camera and projector with respect to the robot end-point can be inferred by moving the robot to a selected set of positions and analyzing the images obtained. The ultimate accuracy of the system is a function of the conditions under which it is evaluated. Presently we estimate the worst-cast accuracy to be @plusminus 5 mm. @i<(14 pages)> @report(title="Overview of ROME: A Reason-Oriented Modeling Environment", author="Donald W. Kosy and Ben P. Wise", number="CMU-RI-TR-85-21", date="December 1985") The ROME system is an experimental decision support system generator that incorporates facilities originally developed for "expert" artificial intelligence systems. Like other DSS generators, it provides a set of tools for building, running, and documenting mathematical models of business activities. Beyond that, however, the system includes explicit differentiation between hard facts and softer information such as assumptions and expectations, a set of procedures for evaluating and explaining results, and a natural language user interface. In this paper, we discuss the representation of knowledge in ROME and show how it is used to answer questions, explain trends, detect anomalous values, and assess the credibility of model assumptions. @i<(12 pages)> @report(title="Model-Based Evaluation of Long-Range Resource Allocation Plans", author="Ben P. Wise and Donald W. Kosy", number="CMU-RI-TR-85-22", date="December 1985") When corporate planning is decentralized, the plans produced by each suborganization must be reviewed and evaluated to make sure they are reasonable and acceptable to the organization as a whole. In this paper we consider three ways of automating the evaluation task: two based on rules combined with qualitative arithmetic, and one based on a microeconomic model combined with quantitative reasoning and a search procedure. We argue that the knowledge encoded in the rules can be represented better using the model and that the search strategy implicit in the rule representation can be duplicated by the procedure. Moreover, quantitative reasoning can deal with reinforcing and counteracting effects, while qualitative arithmetic cannot. This approach has been used as the basis for the REMUS module in the ROME system. @i<(11 pages)>