Date of Award

4-2016

Degree Type

Thesis

Degree Name

Doctor of Philosophy (PhD)

Department

Electrical and Computer Engineering

First Advisor

Jeffrey M. Siskind

Committee Chair

Jeffrey M. Siskind

Committee Member 1

Robert L. Givan

Committee Member 2

Thomas M. Talavage

Committee Member 3

T. N. Vijaykumar

Abstract

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 from training data, and allow actions in new video to be recognized and localized completely automatically. Second, I present a new method which allows a mobile robot to learn word meanings from a combination of robot sensor measurements and sentential descriptions corresponding to a set of robotically driven paths. These word meanings support automatic driving from sentential input, and generation of sentential description of new paths. Finally, I also present work on a new action recognition dataset, and comparisons of the performance of recent methods on this dataset and others.

Download

Share

COinS