This paper describes our current research into nonprehensile palm manipulation. The term “palm ” refers to the use of the entire device surface during manipulation, as opposed to use of the fingertips alone.

The subject of this paper is the analysis of a randomized preprocessing scheme that has been used for query processing in robot path planning. The attractiveness of the scheme stems from its general applicability to virtually any path-planning problem, and its empirically observed success. In this paper we initiate a theoretical basis for explaining this empirical success. Under a simple assumption about the configuration space, we show that it is possible to perform preprocessing following which queries can be answered quickly. En route, we consider related problems on graph connectivity in the evasiveness model, and art-gallery theorems.

This paper presents a new method for computing the configuration-space map of obstacles that is used in motion-planning algorithms. The method derives from the observation that, when the robot is a rigid object that can only translate, the configuration space is a convolution of the workspace and the robot. This convolution is computed with the use of the Fast Fourier Transform (FFT) algorithm. The method is particularly promising for workspaces with many and/or complicated obstacles, or when the shape of the robot is not simple. It is an inherently parallel method that can significantly benefit from existing experience and hardware on the FFT. Keywords: Configuration Space, Fast Fourier Transform, Collision Checking, Robot Motion Planning 1. Introduction The problem of planning the motion of a robot among static obstacles has been recently approached by restricting the configuration space (C-space) of the robot to a discrete space, where each degree of freedom can only assume a fin...

A model is presented for the operation of the striatum. The model posits that the basal ganglia are responsible for driving smooth transitions of state for an organism. We propose that this is accomplished through the computation of a potential function within the striatum on which a gradient descent is performed toward the goal state. The model suggests that various somatotopic regions of the striatum correspond to state spaces, each of which pertains to a different aspect of the organism. This paper discusses this model only in the context of motor control, i.e., egomotion and limb movement. The model appears to account for a variety of experimental results, and for some unusual properties of the striatum. 1 Introduction The basal ganglia have been one of the more enigmatic structures in the brain. While they have been extensively studied (DeLong and Georgopoulos 1981; Wilson, et al. 1990; Kandel, et al. 1991), and have been very clearly implicated in motor function (Yahr 1...

Abstract. One of the goals of humanoid research is the development of a humanoid capable of performing useful tasks in unknown or unpredictable environments. To address the complexities of this task, the robot must continually accumulate and utilize new control and perceptual knowledge. In this paper, we present a control framework for accomplishing this. Robot control policies can be learned at different levels of abstraction. We show how task-relevant perceptual features can be discovered that make better control policies possible. We also explore how trajectories of closed-loop policies can provide uniquely relevant state information. The approach presented in this paper is illustrated with several case studies on actual robot systems. 1

Robotic motion planning requires configuration space exploration. In high-dimensional configuration spaces, a complete exploration is computationally intractable. To devise practical motion planning algorithms in such high-dimensional spaces, computational resources have to be expended in proportion to the local complexity of a configuration space region. We propose a novel motion planning approach that addresses this problem by building an incremental, approximate model of configuration space. The information contained in this model is used to direct computational resources to difficult regions. This effectively addresses the narrow passage problem by adapting the sampling density to the complexity of that region. Each sample of configuration space is guaranteed to maximally improve the accuracy of the model, given the available information. Experimental results indicate that this approach to motion planning results in a significant decrease in the computational time necessary for successful motion planning. 1

Abstract not found

: Analog VLSI provides a convenient and high-performance engine for robot path planning. Laplace's equation is a useful formulation of the path planning problem; however, digital solutions are very expensive. Since high precision is not required an analog approach is attractive. A resistive network can be used to model the robot's domain with various boundary conditions for the source, target, and obstacles. A gradient descent can then be traced through the network by comparing node voltages. We built two analog CMOS VLSI chips to investigate the feasibility of this technique. Design issues included the choice of resistive element, tessellation of the domain, programming of the network and readout of the settled network. Both chips can be connected to a standard VME bus interface to permit their use as co-processors in otherwise digital systems. Keywords: analog VLSI, Laplace's equation, resistive networks, path planning. 1 Introduction Robot path planning is an important area of st...

Collision-free paths for a robot are commonly obtained in configuration space. The major problem with this approach is the computation of the boundary of the configuration space obstacles. While many results have been reported for polygonal environments, the general case of arbitrary object shape has received little attention in the literature so far. This paper presents a new method for tackling this problem in the case of objects for which the boundaries consist of segments of parametrized algebraic curves. A fast numerical algorithm which computes the boundaries of the C-space obstacles in parametrized form is developed. The major result is that initially subdividing the segments guarantees the convergence and limits the computational costs of the applied interval-subdivision Newton method. This makes it possible to compute the exact C--space for this more general class of objects. Finally, two examples with convex objects solved by our implementation are given.

Abstract not found