Graduate Member Spotlight
Expected Graduation Date: Summer 2020
University of Washington
This year, we are excited to have Behnoosh Parsa serve as our Professional Graduate Team Lead, where she will provide resources for our members who are in professional degree programs, and she will provide professional development resources for all our members. Behnoosh worked with other graduate students to begin the GradSWE group at the University of Washington. They work together to provide the graduate students with events that allow for community engagement and stress management. Behnoosh and her colleagues have provided their members with several happy hour events and free yoga sessions during exam times. At the Women in Science and Engineering conference this year, Behnoosh organized a panel on how to create an outstanding CV, and how to translate your research. As an advocate for STEM outreach, Behnoosh has volunteered as a project judge for the FIRST LEGO League Semi-final Competition.
In addition to being a superstar in SWE, Behnoosh is technically well accomplished in her field. Behnoosh has served as a mentor for interns. She ranked 1st in her Ph.D qualifying exam, and won several awards for her research. At the Penn State Graduate Exhibition, she won Best Engineering Research Award. Additionally, she has been recognized with the Kinesiology Research Award. At her institution, she has also been nominated for the Best Teaching Assistant Award.
Research Topic: Machine Learning • Bayesian Learning • Probabilistic Graphical Models • Deep Learning • Reinforcement Learning
Behnoosh is working on two main projects for my Ph.D. The first project is concurrent manipulation of optically actuated micro-robots: Optical tweezers are a device used for manipulating micro-level objects, like cells. The goal of the project is to form a pattern with many microspheres using the optical tweezers. The multi-agent pattern formation is a very involved problem to solve. On the other hand, the environment Behnoosh is dealing with is stochastic due to the Brownian motion, so fast decision making is desirable. Hence, she is using machine learning methods and motion planning and control techniques to design a faster and more reliable solution for this stochastic multi-agent control problem. As a part of this project, Behnoosh developed an algorithm to learn the stochastic dynamics of an optically actuated micro-robot, using a Hierarchical Bayesian Linear Regression model with convergence guarantees which can be found on arxiv.
Behnoosh’s second project involves Amazon Robotics. The goal of the project is to develop an intelligent system able to recognize human actions during object manipulation, and assess the corresponding ergonomic risks. This can be a very important step toward human-robot collaboration, especially if we want the robots to be able to help when they perceive the need.
After graduation, Behnoosh sees herself working as a researcher in a tech company. Outside of work, Behnoosh stays very active! She actively participates in yoga, swimming, weight training, and hiking. She also enjoys baking, and reading positive psychology.