Here is a plain-text order form to order reports by mail. Some of the following reports can be downloaded as PostScript, or compressed PostScript files.
Ralph Hartley, "Propulsion and Guidance in a Simulation of the Worm, C. Elegans," From Animals to Animats 2: Proceedings of the 2nd International Conference on Simulation of Adaptive Behavior, edited by Jean-Arcady Meyer, Herbert L. Roitblat, and Stewart W. Wilson, 122-128, MIT Press, 1993, (NCARAI Report: AIC-93-021). Not available on-line at this time. Please see order form.
Abstract: A simulation of the mechanics of the Nematode worm C. Elegans was constructed and used to study its locomotive behaviors. This simulation shed light on the mechanisms that must be used by any animal or robot using the same type of propulsion and also on aspects of the behavior of the real worm. The simulation of locomotion is an important step in the understanding of the overall behavior of an animal.
Brian Yamauchi, "Dynamical Neural Networks for Mobile Robot Control," Internal Report, 1993, (NCARAI Report: AIC-93-033). Not available on-line at this time. Please see order form.
Abstract: This paper describes research in applying dynamical neural networks (DNNs) to the control of real mobile robots. Previous research had demonstrated the ability of DNNs to control autonomous agents in simulation, and this research was intended to investigate whether these results could be transferred to real world domains. These tasks were explored: predator avoidance, reactive navigation and landmark recognition. For the predator avoidance and landmark recognition tasks, a genetic algorithm was used to evolve DNN controllers in simulation. Then the networks were transferred to the mobile robot for evaluation. In the reactive navigation task, a hand-designed DNN controller was used in combination with algorithms for detecting static and cyclic behavior. In all three of these experiments, the control systems were able to successfully perform the tasks in both simulation and on the real robot.
Harold M. Greenwald and Frank J. Pipitone, "Vcalc: a 3-Space Vector Calculator," Naval Research Laboratory Memorandum Report, NRL/MR/5511--93-7422, December 15,1993, (NCARAI Report: AIC-93-035). Not available on-line at this time. Please see order form.
Abstract: Vcalc is a 3-space vector calculator providing graphic displays for vector and scalar operations. Designed to operate on a Sun Workstation, the program may be run stand-alone or as a debugging tool linked to a C program. It is a straightforward and useful tool whose development was motivated by the difficulty of debugging geometric programs.
Frank Pipitone, "Extracting Elementary Surface Features Using Tripod Operators," Internal Report, December 1993, (NCARAI Report: AIC-93-036). Not available on-line at this time. Please see order form.
Abstract: Tripod operators are a class of feature extraction operators for surfaces. They are useful for recognition and localization based on range or tactile data. They extract a few sparse point samples in a regimented way, so that N sampled surface points yield only N-3 independent scalar features containing all the pose-invariant surface shape information in these points and no other information. They provide a powerful index into sets of prestored surface representations. A tripod operator consists of three points in 3-space fixed at the vertices of an equilateral triangle and a procedure for making several "depth" measurements in the coordinate frame of the triangle, which is placed on the surface like a surveyor's tripod. They have complete six DOF isometry invariance and can be used in many ways and on arbitrary surface shapes. Here we study the structure of tripod feature data on several commonly occurring simple surface types, with the goal of providing concise analytic expressions for mapping sparse surface samples into candidate surface types.
Frank Pipitone and William Adams, "Rapid Recognition of Freeform Objects in Noisy Range Images Using Tripod Operators," 1993 IEEE Conference on Computer Vision and Pattern Recognition, 715-716, IEEE Computer Society Press, June 15-18, 1993, (NCARAI Report: AIC-93-037). Not available on-line at this time. Please see order form.
Abstract: The tripod operator is a class of feature extraction operators for computer-represented surfaces, such as range images. It facilitates the recognition and localization of objects by extracting a few sparse point samples in a regimented way, so that N sampled surface points yield only N-3 independent scalar features containing ALL the surface shape information in these points and NO other information. They have complete six DOF isometry invariance. A tripod operator consists of three points in 3-space fixed at the vertices of an equilateral triangle and a procedure for making several "depth" measurements in the coordinate frame of the triangle, which is placed on the surface like a surveyor's tripod. Tripod operators can be used in many ways for the recognition and/or pose estimation of arbitrary surface shapes. Here we treat the rapid recognition of isolated objects in noisy range images using a statistical treatment of a few isolated tripod operators placements. The objects are initially preprocessed with tripod operators. The resulting feature vector sets are stored. Then new operator placements are made in an image containing an object to be recognized. Experiments show reliable recognition with only a few operator placements, despite significant noise.
Computational Reasoning
Intelligent M4 Systems
Machine Learning
Neural Networks
Back to NCARAI Library