To construct a 3D model of an object in a computer, a range finder is used to obtain range images of the object. The range finder takes range images from many view points to eliminate the influence of occlusions. A registration algorithm is, then, used to merge the range images. When the range finder is mounted on a mobile robot and measures while moving, range images taken by it will be distorted. We extended the registration algorithm to remove the deformation. The extended algorithm can remove it and is used to construct a 3D model even for a case where the speed parameters of the robot are not well known.

A new method is proposed for recognizing 3-D objects by using a Hopfield-style optimization algorithm based on matching between patch-based image and model descriptions. To obtain the image descriptions, range images are employed to extract reliable high-level patch features. In the optimization process the objective function is a Liapunov function that is minimized in a Hopfield net with its interconnections encoding the imposed geometrical constraints on the model descriptions. At first, the paper describes a pre-processing method for deriving the necessary image description. It then presents the structure of the used Hopfield net which is able to recognize multiple objects all at once.

This paper addresses the issue of recognizing 3-D shapes of objects from their 2-D contour pairs. A shape model, that is represented by a family of spheres of varying radii with their centers on its axis, is developed for the shape class of circular generalized cylinders (GCs) with a space-curved axis. Quasi-invariant property of the shape model is defined, and is used to recover the 3-D shapes of the shape class by solving the problem of stereo matching between their contour pairs. The recovered shapes are segmented into their parts and represented by a set of the parametrically describable parts and spatial relations between them, so that they may be used for recognition even under occluded environments.

We constructed a parts picking system that uses two types of sensors; a range finder and a camera system. The parts picking system picks up a part from randomly stacked parts. It uses range data to know the location and the pose of the part to be picked up. It additionally uses the camera system to perform the operation effectively.

This paper presents a method for reasoning volumetric shapes for robotic grasping based on a data-driven approach. The global information about position, orientation and geometrical property of objects is necessary for successful execution of the target approach and the preparation, at the early stage of grasping, rather than fine details of the shapes. A pair of occluding contours is used to extract a simplified 3D shape of circular generalized cylinders (GCs) holding the information. Based on the GC model, which is represented by a family of spheres of varying radii with their centers on its axis, the stereo matching between the contour pair is implemented using the geometrical invariance of spheres. The shapes of blobs and sticks can be simplified as spheres and the families of spheres, respectively. Compared with the range-data based method, its computation is inexpensive, directive and has unique solution. The case of multiple and overlap objects is also addressed.

The authors have proposed a new data structure for an omni-directional view. That is a spherical hexagonal pyramid which has pyramid- and tree-type structures. Such structures manipulate hierarchical and multi-resolution representations of data and it seems useful for mobile robots and manipulators in an assembling task. Because these robots and manipulators have to observe various directions and to utilize image processing for achieving their purposes. This paper describes two methods for observing environment and objects with the aim of confirming of efficiency of this data structure. One is to extract great circles and vanishing points. The information is used for estimating some coordinate systems in an artificial world. And the other is to recognize objects by use of hexagonal chain codes.

This paper presents a new method for cooperative manipulations based on genetic algorithms using contact information. In order to apply genetic algorithms, each robot arm generates new information about object postures using its own contact information without using of any vision systems. The object postures are derived by calculating normal vectors of object surfaces. The system proposed on the basis of genetic algorithms determines forces which robot arms should apply to the objects by solving an optimization problem utilizing the generated and obtained information. Owing to genetic algorithms, the system has not only the adaptation to the change of the working environment, but also can determine the forces for robot arms without physical inconsistency.

Generally, algorithms for assembly task planning are apt to become complicated. Moreover, it is very difficult to create plans for multiple manipulators. The authors propose a new assembly task planning system for multiple manipulators in order to avoid this problem and improve the flexibility and the reliability of the whole system. In this system, each of the parts composing a machine is regarded as an agent, and a plan for an assembly task is automatically generated as the results of autonomous behaviors of these agents. The outline of the planning system based on this concept is described, some simulation results using a cooperative computational model are given.>

For automatic handling of a flexible object such as an electric wire, cable or rope it is necessary to determine the 3-D pose of the object. However, it is difficult to achieve this because of the intrinsic flexibility of the flexible object. We use the k-curvature representation to describe the skeleton image of a flexible object. Moreover, we present two fast stereo matching methods to determine the pose for automatic handling. The one is based on a least square error method and the other is based on the interpolation between curvature extrema. At first, we need to calculate the curvatures of object skeleton images which are taken through two cameras. We apply our algorithms to these two calculated curvature representations. We execute computer simulations to evaluate the validity of the presented algorithms. The results give us some guidelines for the applications with hand-eye robots. Finally, we carry out a camera calibration process for a linear lens model and an experiment to determine the pose of a coaxial cable with the calibrated camera parameters.>

Generally, algorithms for assembly task planning are apt to become complicated. Moreover, it is very difficult to create plans for multiple manipulators, because complicated tasks to divide an obtained plan into some subtasks and allocate them to each manipulator have to be executed. We propose a new assembly task planning system for multiple manipulators in order to avoid these complicated problems and improve the feasibility and the reliability of the whole system. In this system, each of parts composing a machine as regarded as an agent, and a plan for an assembly task is automatically generated as results of autonomous behaviors of these agents. The outline of the planning system based on this concept is described, some simulation results using a cooperative computational model are given.>

In an environment with plural robots, an assembly plan may be changed according to the constraints or abilities of the manipulator which executes the task. For an assembly task composed of subassemblies which can be executed at the same time, the assembly plan depends on the constraints of the manipulator. This paper describes an approach to an assembly planning system using contract net protocols. The contract nets are applied between part agents and robot agents which have a little automatic faculty. An assembly plan is generated based on searching an AND/OR graph for a given product using heuristic estimation.

Recognizing 3-D objects from 2-D images using an aspect graph approach is considered. The authors describe an algorithm for creating aspect graphs for general polyhedral objects, using a closed space centered at the origin of an object as a viewing space. They divide the viewing space into sets of viewpoints according to differences of aspects. An aspect graph is constructed by set operations on sets of viewpoints. The set operations are executed by substituting solid polyhedra for the sets of viewpoints. The algorithm enables the authors to classify aspects of a general polyhedral object under perspective projection, to solve the problem of occlusion caused when an object has nonconvex parts, and to construct aspect graphs automatically.

A method for quantifying saliency of 3D feature points from range images of model surfaces is proposed. The saliency is characterized by the discriminating power of the feature points for distinguishing the model objects they belong to from others. It is measured by local shape similarity, and the similarity coefficients of surface points are calculated by parameterizing local surface representations into the local orientation coordinate systems. An algorithm for performing the Hough transform weighted by the similarity coefficients is developed to optimally determine the saliency coefficients of the feature points. It is shown that the method is especially useful for dealing with the occlusion problem that must be solved when constructing a flexible robot vision system.>

We propose a method to solve the colli- sion free motion planning problem under the situa- tion where multiple manipulators in the same space are engaged in works in an independent or a loosely cooperative manner. The strategy is based on the concept of the autonomous and cooperative dis- tributed robots system. First of all, a basic database utilized for mapping from a work space(W-space) to a configuration space(C-space) and managed in or- der to construct obstacle areas in C-space easily and rapidly is described. Second, a correlation map between C-space and W-space is introduced in order to define an area, where each manipulator has the possibility to col- lide each other, in W-space on the basis of reachable areas of each manipulator in C-space. Third, considered as a resource allocation prob- lem in W-space, the collision free path problem for each manipulator is solved in the frame of the cooperative distributed problem-solving for an au- tonomous distributed system. The approach is simi- lar to the cooperation strategy for avoiding the con- flict of airplanes in an air-traffic control domain. Simulation results for a two manipulators system re- veal the usefulness of the method we proposed.

A way of parameterizing a spherical obstacle in the configuration space (C-Space) and the properties of a moving spherical obstacle are described. A spherical obstacle in a work space (W-Space) is transformed into a characteristic region in C-Space, and can be approximated as a set of geometric objects. The transformation of a complex work space into C-Space is described in terms of unifying the transformation of each element of a set of spheres. A collision-free path is made by using a penalty-function-method. for a moving spherical obstacle in the work space, the penalty-function is modified according to the factors of the velocity along axes in polar coordinates system. Simulation results are presented for the case of two manipulators in W-Space.>

An approach to recognizing and locating a partially visible object from multiple images for a pile of parts is proposed. The image-to-model correspondence is established through an examination of the consistency between the surface patches extracted from the images and the patches described in an object model. To obtain the scene description, which is composed of parameterized surface patches, the scene's needle map is derived by a modified photometric stereo method. The needle map is then segmented into primitive surfaces, and the resultant patches are identified and described by using a modified Hough transformation based on the Gaussian spherical maps of the patches. To interpret the scene description, the object model is built with a frame structure to describe geometric features of the model patches. By a search process guided by some heuristic planning strategies based on the fuzzy-set concept, sets of patches considered to be possibly on the surface of the model object are extracted from the scene description. The candidate sets are tested and the locations and orientations of the corresponding object instances are computed. Results of two experiments are reported.>

A robot plan-generation system is described that treats continuous state changes over time for multiple robots. A model for a continuous domain is represented, and a parallel plant-generation system, based on production rules for multiple robots in this domain, is proposed. The system consists of a fundamental planning subsystem for multiple robots and a subsystem for detecting and avoiding mutual collisions of cylindrical-type robots, called PLAMAT and SYDAMUC, respectively. In addition to examples for each subsystem, an assembly problem is solved as an example for the total plan-generation system, and the usefulness of the system is confirmed.>