Showing papers by "Avinash C. Kak published in 1986"

Determination of the identity, position and orientation of the topmost object in a pile

Abstract: In this paper, we first propose schemes for segmenting out the visible part of the topmost object from a pile of planar and curved objects. We then describe our work on using B -splines for the characterization of the topmost object surface when it is curved; the B -splines are used for deriving operators that yield the Gaussian and mean curvatures. This is followed by a description of our identification strategies which depend upon whether the topmost object is judged to be planar or curved. The identification strategy for planar objects revolves around the EGI representation of their visible surfaces and is a function of whether the number of visible planar surfaces is one, two, three, or more. In case the number of sufficiently visible planar surfaces is less than or equal to two, we have incorporated surface boundary information with angular relation between adjoining surfaces to improve the identification process. For curved objects, the identification strategy depends upon the signs of the Gaussian and the mean curvatures and the EGI. While these identification strategies are not guaranteed to work in every case, we expect them to be practically useful for a wide range of industrial objects.

Determination of the identity, position and orientation of the topmost object in a pile: Some further experiments

Abstract: In a recent report [7], we proposed algorithms for segmenting out the visible part of the topmost object from a pile of planar and curved objects. Planar objects considered were of convex polyhedral type, such as prisms, boxes, wedges, etc; and the curved objects were of the type that could be recognized uniquely by using the Extended Gaussian Image, such object-types being cylinders, cones, spheres, ellipsoids, toruses, etc. For input these algorithms used 3-D vision data acquired with a structured light scanner. In this report, we will review some of the algorithms presented earlier and show our latest experimental results on scenes that are more complex than before. In our presentation at the conference, we also plan to show some manipulation experiments that for sensory feedback use 3-D vision data analyzed according to the algorithms presented here.

FProlog: A language to integrate logic and functional programming for automated assembly

Abstract: In this paper, we present FProlog, a programming language designed to act as the top level in a robot assembly system. FProlog is a logic programming language, with the ability to interface with LISP. This allows the use of a logic programming environment to construct assembly plans, while using LISP programs to interface with vision systems, world modeling systems, robot manipulators, etc. FProlog differs from hybrid logic programming languages, such as LOGLISP, in that FProlog may invoke functional programs as goals, and functional programs may invoke FProlog's inference engine. Also, FProlog differs from traditional robot assembly languages, such as AUTOPASS, in its generality, and therefore its ability to interface with many different subsystems. As a demonstration of the applicability of FProlog, we also present an FProlog program which is used as the top level in a robot assembly system which performs a version of the blocks world experiment.

Knowledge-Based Stereo and Structured Light for 3-D Robot Vision

Abstract: In this report, we first show that a knowledge based stereo using a single Marr-Poggio-Grimson (MPG) channel is as effective for range mapping as a stereo that uses no a priori knowledge and all five channels. For this comparison, the maximum allowable depth of the scene in both cases is seven times what is matchable by the disparity range of the single channel in the knowledge-based system, which is also the finest channel for the full MPG implementation. To invoke the knowledge base, prior to the application of the MPG channel, the identification of the object class and approximate determination of the object location are performed using a novel structured-list data base extracted from the radial-valued skeletal image. World knowledge and object models are represented as LISP frames. The second half of the report which deals with structured light for 3-D vision contains two innovations: 1) We have presented a new scheme for extracting edges from range data, which consists of first constructing a synthetic image from the range data in which the gray level at each point is proportional to the orientation of the local surface normal and then applying a reflectance data edge detector to this synthetic image; and 2) we have briefly presented our PILE ANALYSIS program, which is a high level symbolic processing program for robot vision. For a scene consisting of a pile of overlapping polyhedral objects, the pile analysis program uses the range data to isolate the edges and the vertices of the objects at the top of the heap. In this half of the report, we have also discussed binary encoded and proposed color encoded parallel grid projections for improved accuracy and speed in range data acquisition. Color encoded parallel grid projection offers the possibility of real-time range data acquisition at high resolution for modest cost.