Talking Robotics

Logo


Organizers: Patrícia Alves-Oliveira,
Silvia Tulli, Miguel Vasco
contact us: talkingrobotics at gmail dot com — support us: buymeacoffe

twitter youtube

Speaker

Marynel Vázquez is an Assistant Professor in Yale’s Computer Science Department, where she leads the Interactive Machines Group. Her research focuses on Human-Robot Interaction (HRI), specially in multi-party settings. Marynel and her students investigate social group phenomena in HRI and develop perception and decision making algorithms to enable autonomous robot behavior. Marynel received her bachelor’s degree in Computer Engineering from Universidad Simón Bolívar in 2008, and obtained her M.S. and Ph.D. in Robotics from Carnegie Mellon University in 2013 and 2017, respectively. Before joining Yale, Marynel was a collaborator of Disney Research and a Post-Doctoral Scholar at the Stanford Vision & Learning Lab.

Speaker Links: Website - Google Scholar - Github - Lab Website - Twitter


Abstract

Over the past decade, machine learning has led to tremendous progress in many research areas, like computer vision, natural language processing and robot manipulation. Oftentimes, these advances have been attributed to two factors: (1) the availability of increased amounts of data, and (2) advancements in deep learning. However, the same level of impact has not yet been observed in Human-Robot Interaction (HRI). One important reason is that data collection in HRI, which is often achieved via user experiments, can be expensive and very time consuming. Also, properties of many HRI settings, especially those involving multi-party interactions, pose additional challenges for applied machine learning. For example, one may need to reason not only about individual users in HRI, who may leave and join interactions, but also about group factors that may drive human behavior.

In this talk, I will discuss how recent advancements in deep learning could aid in tackling important challenges in multi-party HRI. In particular, I will advocate for Graph Neural Networks as an important class of models that can open up opportunities for reasoning about individual, inter-personal and group-level factors in human-robot interactions. During the talk, I will also discuss approaches to overcome the data scarcity problem. I will present an idea on how the HRI community could more easily collect data for early system development and testing via interactive online surveys. We have begun to explore this idea in the context of social robot navigation but, thanks to advances in game development engines, it could be easily applied to other application domains.


Papers covered during the talk