RPI, Graduate Student Presentations

 

RPI, Graduate Student Presentations

Sage 4101

April 15, 2009 12:00 PM - 1:30 PM

Jon Matthis:

Title: The role of sensory information in self-judgments of action capabilities

Abstract: Navigating through dynamic environments requires actors to perceive and act in ways that take into account their locomotor capabilities. For instance, to determine whether a shrinking gap is passable, an actor must consider their movement potential relative to key features of their environment. When making judgments about their action capabilities, actors may rely on what they know about their capabilities based on past experience, or they may utilize sensory information picked up "on the fly." We conducted two experiments in a virtual environment that required subjects to make judgments about their ability pass through a pair of converging posts. These experiments manipulated the visual information available to subjects during this task in order to examine the way they used sensory information to make judgments about their action capabilities.

Romann M. Weber:

Title: "Nature's Tread: The Evolutionary Benefit of Water-Immersion Skin Wrinkling in Primates" Romann Weber)

Abstract: Water-immersion skin wrinkling on the hands and feet is an involuntary, neurally mediated response to the sensing of water. This response seems to demonstrate an unconscious "awareness" of environmental conditions and their possible effects on the body. In this talk, I will discuss the hypothesis that this effect is a selected-for modification that improves animals' ability to negotiate wet surfaces. Preliminary data will show that this response significantly improves grip in wet conditions. The qualitative similarities between the patterns found in water-immersion skin wrinkling and those found in natural drainage networks will also be discussed.

Michel Brudzinski:

Title: Visual Similarity is ObviS

Abstract: The ObViS (pronounced like obvious) research project involves the development and validation of a computational measure of visual similarity. We have adapted a previously published computer vision algorithm to the task of concisely representing an image as a vector. Vector representations for images can then be compared to measure similarity. We have conducted a series of experiments to validate this algorithm as a measure of visual similarity, including both implicit and explicit human judgments of visual similarity. Our results suggest that the Obvis algorithm could be a useful measure of visual similarity. This algorithm could have applications in visual search modeling and user interface design.

Jason Ralph:

To be announced.

Add to calendar
Share|