Artificial Intelligence and Robotics Commons^™

Classifying Pattern Formation In Materials Via Machine Learning, Lukasz Burzawa, Shuo Liu, Erica W. Carlson

The Summer Undergraduate Research Fellowship (SURF) Symposium

Scanning probe experiments such as scanning tunneling microscopy (STM) and atomic force microscopy (AFM) on strongly correlated materials often reveal complex pattern formation that occurs on multiple length scales. We have shown in two disparate correlated materials that the pattern formation is driven by proximity to a disorder-driven critical point. We developed new analysis concepts and techniques that relate the observed pattern formation to critical exponents by analyzing the geometry and statistics of clusters observed in these experiments and converting that information into critical exponents. Machine learning algorithms can be helpful correlating data from scanning probe experiments to theoretical models …

Interactive Logical Analysis Of Planning Domains, Rajesh Kalyanam

Open Access Dissertations

Humans exhibit a significant ability to answer a wide range of questions about previously unencountered planning domains, and leverage this ability to construct “general-purpose'' solution plans for the domain.

The long term vision of this research is to automate this ability, constructing a system that utilizes reasoning to automatically verify claims about a planning domain. The system would use this ability to automatically construct and verify a generalized plan to solve any planning problem in the domain. The goal of this thesis is to start with baseline results from the interactive verification of claims about planning domains and develop the …

Data Driven Low-Bandwidth Intelligent Control Of A Jet Engine Combustor, Nathan L. Toner

Open Access Dissertations

This thesis introduces a low-bandwidth control architecture for navigating the input space of an un-modeled combustor system between desired operating conditions while avoiding regions of instability and blow-out. An experimental procedure is discussed for identifying regions of instability and gathering sufficient data to build a data-driven model of the system's operating modes. Regions of instability and blow-out are identified experimentally and a data-driven operating point classifier is designed. This classifier acts as a map of the operating space of the combustor, indicating regions in which the flame is in a "good" or "bad" operating mode. A data-driven predictor is also …

Texture Modelling Using Convolutional Neural Networks, Leon A. Gatys, Alexander S. Ecker, Matthias Bethge

MODVIS Workshop

We introduce a new model of natural textures based on the feature spaces of convolutional neural networks optimised for object recognition. Samples from the model are of high perceptual quality demonstrating the generative power of neural networks trained in a purely discriminative fashion. Within the model, textures are represented by the correlations between feature maps in several layers of the network. We show that across layers the texture representations increasingly capture the statistical properties of natural images while making object information more and more explicit. Extending this framework to texture transfer, we introduce A Neural Algorithm of Artistic Style that …

Focusing On Selection For Fixation, John K. Tsotsos, Calden Wloka, Yulia Kotseruba

MODVIS Workshop

Building on our presentation at MODVIS 2015, we continue in our quest to discover a functional, computational, explanation of the relationship among visual attention, interpretation of visual stimuli, and eye movements, and how these produce visual behavior. Here, we focus on one component, how selection is accomplished for the next fixation. The popularity of saliency map models drives the inference that this is solved; we suggested otherwise at MODVIS 2015. Here, we provide additional empirical and theoretical arguments. We then develop arguments that a cluster of complementary, conspicuity representations drive selection, modulated by task goals and history, leading to a …

Grounding Robot Motion In Natural Language And Visual Perception, Scott Alan Bronikowski

Open Access Dissertations

The current state of the art in military and first responder ground robots involves heavy physical and cognitive burdens on the human operator while taking little to no advantage of the potential autonomy of robotic technology. The robots currently in use are rugged remote-controlled vehicles. Their interaction modalities, usually utilizing a game controller connected to a computer, require a dedicated operator who has limited capacity for other tasks.

I present research which aims to ease these burdens by incorporating multiple modes of robotic sensing into a system which allows humans to interact with robots through a natural-language interface. I conduct …

Learning In Vision And Robotics, Daniel P. Barrett

Open Access Dissertations

I present my work on learning from video and robotic input. This is an important problem, with numerous potential applications. The use of machine learning makes it possible to obtain models which can handle noise and variation without explicitly programming them. It also raises the possibility of robots which can interact more seamlessly with humans rather than only exhibiting hard-coded behaviors. I will present my work in two areas: video action recognition, and robot navigation. First, I present a video action recognition method which represents actions in video by sequences of retinotopic appearance and motion detectors, learns such models automatically …

On The 3d Point Cloud For Human-Pose Estimation, Kai-Chi Chan

Open Access Dissertations

This thesis aims at investigating methodologies for estimating a human pose from a 3D point cloud that is captured by a static depth sensor. Human-pose estimation (HPE) is important for a range of applications, such as human-robot interaction, healthcare, surveillance, and so forth. Yet, HPE is challenging because of the uncertainty in sensor measurements and the complexity of human poses. In this research, we focus on addressing challenges related to two crucial components in the estimation process, namely, human-pose feature extraction and human-pose modeling.

In feature extraction, the main challenge involves reducing feature ambiguity. We propose a 3D-point-cloud feature called …

Video Event Understanding With Pattern Theory, Fillipe Souza, Sudeep Sarkar, Anuj Srivastava, Jingyong Su

MODVIS Workshop

We propose a combinatorial approach built on Grenander’s pattern theory to generate semantic interpretations of video events of human activities. The basic units of representations, termed generators, are linked with each other using pairwise connections, termed bonds, that satisfy predefined relations. Different generators are specified for different levels, from (image) features at the bottom level to (human) actions at the highest, providing a rich representation of items in a scene. The resulting configurations of connected generators provide scene interpretations; the inference goal is to parse given video data and generate high-probability configurations. The probabilistic structures are imposed using energies that …

Two Correspondence Problems Easier Than One, Aaron Michaux, Zygmunt Pizlo

MODVIS Workshop

Computer vision research rarely makes use of symmetry in stereo reconstruction despite its established importance in perceptual psychology. Such stereo reconstructions produce visually satisfying figures with precisely located points and lines, even when input images have low or moderate resolution. However, because few invariants exist, there are no known general approaches to solving symmetry correspondence on real images. The problem is significantly easier when combined with the binocular correspondence problem, because each correspondence problem provides strong non-overlapping constraints on the solution space. We demonstrate a system that leverages these constraints to produce accurate stereo models from pairs of binocular images …

Formal Aspects Of Non-Rigid-Shape-From-Motion Perception, Vicky Froyen, Qasim Zaidi

MODVIS Workshop

Our world is full of objects that deform over time, for example animals, trees and clouds. Yet, the human visual system seems to readily disentangle object motions from non-rigid deformations, in order to categorize objects, recognize the nature of actions such as running or jumping, and even to infer intentions. A large body of experimental work has been devoted to extracting rigid structure from motion, but there is little experimental work on the perception of non-rigid 3-D shapes from motion (e.g. Jain, 2011). Similarly, until recently, almost all formal work had concentrated on the rigid case. In the last fifteen …

Object Recognition And Visual Search With A Physiologically Grounded Model Of Visual Attention, Frederik Beuth, Fred H. Hamker

MODVIS Workshop

Visual attention models can explain a rich set of physiological data (Reynolds & Heeger, 2009, Neuron), but can rarely link these findings to real-world tasks. Here, we would like to narrow this gap with a novel, physiologically grounded model of visual attention by demonstrating its objects recognition abilities in noisy scenes.

To base the model on physiological data, we used a recently developed microcircuit model of visual attention (Beuth & Hamker, in revision, Vision Res) which explains a large set of attention experiments, e.g. biased competition, modulation of contrast response functions, tuning curves, and surround suppression. Objects are represented by …

Modeling Visual Features To Recognize Biological Motion: A Developmental Approach, Giulio Sandini, Nicoletta Noceti, Alessia Vignolo, Alessandra Sciutti, Francesco Rea, Alessandro Verri, Francesca Odone

MODVIS Workshop

In this work we deal with the problem of designing and developing computational vision models – comparable to the early stages of the human development – using coarse low-level information.

More specifically, we consider a binary classification setting to characterize biological movements with respect to non-biological dynamic events. To this purpose, our model builds on top of the optical flow estimation, and abstract the representation to simulate the limited amount of visual information available at birth. We take inspiration from known biological motion regularities explained by the Two-Thirds Power Law, and design a motion representation that includes different low-level features, …

Trajectory Generation For Lane-Change Maneuver Of Autonomous Vehicles, Ashesh Goswami

Open Access Theses

Lane-change maneuver is one of the most thoroughly investigated automatic driving operations that can be used by an autonomous self-driving vehicle as a primitive for performing more complex operations like merging, entering/exiting highways or overtaking another vehicle. This thesis focuses on two coherent problems that are associated with the trajectory generation for lane-change maneuvers of autonomous vehicles in a highway scenario: (i) an effective velocity estimation of neighboring vehicles under different road scenarios involving linear and curvilinear motion of the vehicles, and (ii) trajectory generation based on the estimated velocities of neighboring vehicles for safe operation of self-driving cars during …

Securing Communication Within The Harms Model For Use With Firefighting Robots, Maxwell D. Dewees

Open Access Theses

Humans and robots must work together in increasingly complex networks to achieve a common goal. In this research, firefighting robots are a part of a larger, decentralized system of humans, agents, robots, machines, and sensors (HARMS). Although communication in a HARMS model has been utilized in previous research, this new study looks at the security considerations of the communications layer of the HARMS model. A network attack known as a man-in-the-middle attack is successfully demonstrated in this paper. Then, a secure communications protocol is proposed to help provide confidentiality and authentication of HARMS actors. This research is applied to any …

Improving Product Design Phase For Engineer To Order (Eto) Product With Knowledge Base Engineering (Kbe), Hanhdung Thi Dinh

Open Access Theses

In industry currently Computer Aided Design (CAD) is an important tool for the modification, analysis, or optimization of the 3D virtual environment that replicates the physical product. CAD software is an efficient and reliable tool. However, as globalization increases customer demands, this process needs to be faster and more efficient to accommodate changing product design situations, especially for Engineer-to- Order (ETO) products. ^ The traditional method of product design process is to operate CAD software without argumentation. Design engineers create CAD prototypes and drawings based on available knowledge and information which comes from engineering experts, company standards, industrial practices as …

Improved Microrobotic Control Through Image Processing And Automated Hardware Interfacing, Archit R. Aggarwal, Wuming Jing, David J. Cappelleri

The Summer Undergraduate Research Fellowship (SURF) Symposium

Untethered submilliliter-sized robots (microrobots) are showing potential use in different industrial, manufacturing and medical applications. A particular type of these microrobots, magnetic robots, have shown improved performance in power and control capabilities compared to the other thermal and electrostatic based robots. However, the magnetic robot designs have not been assessed in a robust manner to understand the degree of control in different environments and their application feasibility. This research project seeks to develop a custom control software interface to provide a holistic tool for researchers to evaluate the microrobotic performance through advance control features. The software deliverable involved two main …

Moving Object Detection For Interception By A Humanoid Robot, Saltanat B. Tazhibayeva

Open Access Theses

Interception of a moving object with an autonomous robot is an important problem in robotics. It has various application areas, such as in an industrial setting where products on a conveyor would be picked up by a robotic arm, in the military to halt intruders, in robotic soccer (where the robots try to get to the moving ball and try to block an opponent's attempt to pass the ball), and in other challenging situations. Interception, in and of itself, is a complex task that demands a system with target recognition capability, proper navigation and actuation toward the moving target. There …