William E. Marsh

VirtuTrace

William Marsh — Sat, 31 Dec 2011 20:08:51 +0000

VirtuTrace is a virtual reality experiment engine that was originally designed for studying firefighter decision making. It has been completely refactored with the intention of it becoming a more general-purpose engine, allowing an experimenter to quickly create virtual worlds based on 3D models and easily mix and match these worlds with different navigation interfaces, physics properties, and scene logic.

In particular, I have been responsible for developing the physics and navigation components of VirtuTrace. Real-world physics is currently implemented using Bullet Physics (with help from osgBullet). The following navigation interfaces have been implemented:

BodyNavigation – This is a hybrid interface utilizing both position- and rate-control to allow for semi-natural locomotion from within the bounds of a six-sided CAVE.
WiiSegwayNavigation – This interface incorporates a Wii Remote and Wii Balance Board for locomotion similar to using a Segway.
GamepadNavigation – This interface incorporates a standard two-stick gamepad with one stick mapped to virtual translation and the other mapped to virtual rotation.

Collaborators
Dr. Nir Keren
Ross Bohner
Kevin Godby

Virtual Locomotion

William Marsh — Fri, 27 Aug 2010 00:17:44 +0000

This is the research featured in my dissertation. Users often wish to navigate through an infinite virtual world from within the confines of a virtual reality (VR) system. In some cases these confines are physical (CAVE walls) and in others the range of tracking hardware is the restriction. This invariably leads to the use of less-than-natural locomotion interfaces for infinite virtual spaces. In addition to constraints on input techniques, VR systems also fail to provide high-fidelity sensory feedback when moving through the virtual world. I am interested in exploring the specific cognitive implications of these design choices, particularly with respect to use of working memory resources.

I have been using Dr. Nir Keren’s VirtuTrace application as a testbed for locomotion interfaces. As part of this, I have been helping with a major refactoring project, focusing on the navigation and physics portions of the code. In the summer of 2010, I was the graduate student mentor for a Research Experience for Undergraduates (REU) team. The team helped with the creation of a locomotion interface that uses a WiiMote and Wii Balance Board to simulate riding a Segway. They also conducted a pilot study in which this interface was compared with another, body-based interface.

My first study investigated the bi-directional impact of specific working memory tasks on individual locomotion movements using three different interfaces. Results show that interfaces that are less natural require additional spatial working memory resources, as opposed to verbal or general attention resources.

The second study involved similar tasks but it compared high field of view to low field of view. Results indicate that additional general attention resources are required when the field of view is reduced.

Based on the results of the first two studies, an adaptive system (fuzzy inference system) was created to adjust interface parameters according to the number of items that a user was currently being asked to remember. A study showed that this interface improves stopping performance and reduces virtual collisions.

Collaborators
Dr. Jim Oliver
Dr. Jon Kelly
Dr. Veronica Dark
Dr. Les Miller
Dr. Julie Dickerson
Ross Bohner
Kevin Godby

Resources
IEEE VR 2012 short paper (coming soon)
SPIE The Engineering Reality of Virtual Reality 2012 paper
HCII 2011 poster
HCII 2011 poster abstract (The original publication is available at www.springerlink.com.)
ETC 2011 poster

Displays for UAV Path Planning

William Marsh — Tue, 27 Jul 2010 00:40:23 +0000

In command and control of unmanned aerial vehicles (UAVs), increased autonomy allows for vehicles to fly and perform mission objectives largely without operator input. However, when unexpected circumstances arise the interface must allow for rapid and effective re-tasking. Such an interface must display 3-dimensional path information so it can be quickly understood. In these studies, we compared a top-down 2D display with a 3D-perspective display for facilitating an operator’s ability to effectively re-route unmanned aerial vehicles on the battlefield. Additionally, we looked at the pros and cons of providing an interactive display as compared to auto-panning.

Collaborators
Levi Swartzentruber
Joe Holub
Dr. Jim Oliver
Dr. Eliot Winer
Dr. Stephen Gilbert

Resources
WINVR 2010 paper – Interfaces for 3D Flight Path Visualization

Incongruent Movement

William Marsh — Sun, 27 Jun 2010 00:38:17 +0000

This stemmed from a class project involving human sense of movement. We were interested in altering visual feedback such that the effect of stepping was opposite, slower, or faster than in the real world. Our team developed an “incongruent” testbed application in which users were required to hit a virtual ball with their body.

Collaborators
Daniela Faas
Kofi Whitney
David Niedergeses

Edible Earth

William Marsh — Sat, 29 May 2010 00:47:07 +0000

College students tend to choose food that is cheap and convenient and they often overestimate the amount of time and difficulty required to find and prepare locally-grown food. We present a conceptual location-aware mobile application intended to help users find local and seasonal ingredients and then prepare meals using them. This could help reduce the environmental impact of their produce choices and encourage healthier eating. This project was among ten selected for the CHI Student Design Competition poster session.

Collaborators
Ross Bohner
Adam Faeth
Nikki D’Adamo
Sara Kaplan

Resources
CHI 2009 Student Design Competition extended abstract