A. M. Erkmen

Information fractals for evidential pattern classification

Proposed is a novel model of belief functions based on fractal theory. The model is first justified in qualitative, intuitive terms, then formally defined. Also, the application of the model to the design of an evidential classifier is described. The proposed classification scheme is illustrated by a simple example dealing with robot sensing. The approach followed is motivated by applications to the design of intelligent systems, such as sensor-based dexterous manipulators, that must operate in unstructured, highly uncertain environments. Sensory data are assumed to be (1) incomplete and (2) gathered at multiple levels of resolution. >

Entropy-driven on-line control of autonomous robots

A new approach to on-line path planning is derived in this paper. The planning algorithm is motivated by robot navigation and manipulation tasks in uncertain, unstructured, dynamic environments. A minimum entropy evidential classifier is used to recognize targets and obstacles in the environment. An iterative Newton scheme is then used to generate a sequence of knot points that guide the motion of the robot. The acquisition and processing of sensory data continue during the motion, thus reducing the uncertainty about the environment. The classification of targets and obstacles is updated, and the path is replanned (locally) to adapt to those changes. A graphical tool based on the concept of Julia sets is used to ensure the predictability and smoothness of the paths.

Online path planning under uncertainty

The authors deal with an online planning algorithm. The work is motivated by robot navigation and manipulation tasks in unstructured, dynamic environments. It is assumed that sensory information is incomplete and must be expanded and/or redefined by active sensing during an exploratory motion phase. Candidate targets are modeled as attractors, while obstacles are modeled as repellers. Path planning is reduced to an iterative Newton scheme that can readily adapt to changes in the environment and to new sensory information. Julia sets are used to detect and avoid chaotic convergence.<<ETX>>

Preshape Jacobians for minimum momentum grasping

Dexterous grasps for multifingered robot hands are planned. A real-time, joint space finger path planning algorithm is derived for the enclosure phase of grasping motion. The algorithm minimizes the impact momentum of the hand. It uses a preshape Jacobian matrix to map task-level hand preshape requirements into kinematic constraints. A master-slave scheme avoids inter-finger collisions and reduces the dimensionality of the planning problem.<<ETX>>

Information fractals in evidential reasoning

Evidential reasoning based on a fractal model of belief is outlined. The specific focus is on the fractal modeling of belief functions. After a qualitative justification and interpretation of this model, several concepts and tools needed for its incorporation into evidential reasoning are formally defined. A particularly important concept is that of conductivity, as it provides the basis of partial evidential matching in the present approach to reasoning by analogy. A conductivity analysis algorithm is derived, and it is illustrated by an application to a simple object classification problem. The fractal model provides potentially powerful mechanisms for a quantitative measure of relevance of a piece of evidence to a knowledge base, and a systematic approach to the coarsening and refining of frames of discernment. The proposed model is motivated by applications to the design of intelligent systems, such as sensor-based dexterous manipulators that must operate in unstructured environments in the presence of high levels of uncertainty.<<ETX>>

Model‐based sensory pattern fusion

This article describes an evidential pattern classifier for the combination of data from physically different sensors. We assume that the sensory evidence is multiresolutional, incomplete, imprecise, and possibly inconsistent. Our focus is on two types of sensory information patterns: visual and tacticle. We develop a logical sensing scheme by using a model-based representation of prototypical 3-D surfaces. Each surface represents a class of topological patterns described by shape and curvature features. The sensory evidence is classified by using a conductivity measure to determine which prototypical surface best matches the evidence.

Shape and Curvature Data Fusion by Conductivity Analysis

This work deals with the design of a new approach to the fusion of sparse or incomplete sensory data, as required by intelligent robot systems in unstructured, highly uncertain environments.

Multiresolutional Sensor Fusion By Conductivity Analysis

This paper describes an evidential pattern classifier for the combination of data from physically different sensors. We assume that the sensory evidence is multiresolutional, incomplete, imprecise, and possibly inconsistent. Our focus is on two types of sensory information patterns: visual and tactile. We develop a logical sensing scheme by using a model based representation of prototypical 3D surfaces. Each surface represents a class of topological patterns described by shape and curvature features. The sensory evidence is classified by using a conductivity measure to determine which prototypical surface best matches the evidence. A formal evidential model of uncertainty is used to derive logical sensors and provide performance measures for sensor integration algorithms.