Research
and Publications
Current Work: Developing Assistive Robots for the Visually Impaired - How does your intended demographic influence information fusion of communications from a robot?
This work is currently in progress and will be updated later. I am looking for participants in my anonymous, asynchronous, online study! Compensation is a $15 Amazon Giftcard sent to the email address of your choice. You can participate in the study (while it's still open!) here. If you have any questions, feel free to email me at Audrey.Balaska@tufts.edu.
Modeling Human/Robot Search Teams and Evaluating Performance in Simulation
At Tufts University in the Automated Systems and Robotics (ASAR) lab, we explored using agent-based modeling approaches to model team search with uncertain detection, and the total number of hidden objects unknown (such as would be the case in search and rescue operations). With this model, we created a Bayesian Estimation Method to estimate the total number of objects, N, hidden in the search area (including objects not yet found) and estimated the searcher skill level, S. In more recent work, we also started modeling how this would impact the ability to monitor information conflict; specifically, we developed a time-sensitive conflict threshold to identify acceptable (and expected) amounts of information differences.
Slide from ION ITM Presentation on Bayesian Estimation Algorithm
Upper plot shows that the searcher has a relatively high skill level, and the lower estimates that there are roughly 10 objects hidden, despite only 8 being found.
Image from IEEE HST 2022 Paper
Plot shows a threshold for information conflict over time. The conflict scores are from a case where Agent A (of A, B, and C) had wrong background information. The grey lines falling outside of the bounds of the black threshold shows that there is persistent information conflict.
Photo Credit: Jeremy Gasowski
Undergraduate Research - Autonomous Surface Vehicle @ University of New Hampshire
From June 2018 to May 2019, I worked on the Autonomous Surface Vehicle (ASV) project, first as a summer undergraduate researcher and then as part of my senior capstone project. The goal of the project was autonomous ocean mapping using pairs of ASVs and Unmanned Underwater Vehicles (UUVs). I worked on the theoretical control design of the ASV; both our larger system (shown in the figure on the left) and in our smaller prototype for tank testing. In addition, I served as the team lead of the interdisciplinary team, which required writing grant proposals through UNH programs, representing our team at community events, and assisting teammates as needed (including doing engineering design, helping with testing of our system, and teaching control theory to computer science students and underclassmen).
Image taken from documentation. Note, I modified the electronics, the casing was kept from the older prototype.
NSF REU - Rehabilitation Engineering @ Cleveland State University
Summer 2017
I was part of the first year cohort for the Rehabilitation Engineering National Science Foundation (NSF) Research Experiences for Undergraduates (REU) site at Cleveland State University. As a rising junior, I was responsible for reverse engineering a non-functioning wireless load cell prototype, troubleshooting what did/didn't work, and then 46 pages documenting the updated version.