David Porfirio

Human-Computer Interaction Lab • University of Wisconsin–Madison



portrait I'm a fourth-year PhD student at the University of Wisconsin–Madison, working on social robotics and formal methods. Specifically, I create tools and techniques that help people program social robots. I work with professors Bilge Mutlu, Aws Albarghouthi, and Allison Sauppé, and I am currently supported by the NSF Graduate Research Fellowship.

Prior to my graduate studies, I attended the University of Arizona for my undergraduate degree. As an undergraduate I worked with Dr. E. Fiona Bailey in the Physiology Department to record and analyze neuromuscular activity, Dr. Joanna Masel in the Ecology and Evolutionary Biology Department to assist developing computational models of evolution, and Dr. John Kececioglu in the Computer Science Department to develop a novel method for protein secondary structure prediction.

You can download my CV here.


Projects

(starting from the most recent)

Programatic Repair of Human-Robot Interactions

We are currently investigating novel methods for automatically making modifications to a robot program after the program has been deployed on a real robot. The goal of the modifications is to maximize user experience for a specific interaction context, while maintaining adherence to a prespecified set of baseline context-free social norms. Thus, our work necessitates combining optimization with model checking.
portrait

Bodystorming Human-Robot Interactions

We developed a programming environment, Synthé, that enables design teams to act out, or bodystorm, human robot interactions. Designer demonstrations are converted to execution traces, which are then used as input to an inductive synthesis algorithm which synthesizes a full human-robot interaction program from scratch. Synthé's source code will be available on github soon.
portrait

Authoring and Verifying Human-Robot Interactions

We developed a visual programming environment that allows people to design human-robot interaction programs and receive feedback in real-time on whether these programs violate social norms. In order to provide this feedback, we model in-progress human-robot interaction programs as transition systems, and a set of context-specific social norms within temporal logic. The transition systems and social norms are then input into an off-the-shelf model checker. RoVer's source code is located at https://github.com/Wisc-HCI/RoVer.

dporfirio{at}wisc.edu