Pat Callaghan

Hi! I’m Pat, and I’m a PhD student in the Robotics Institute at Carnegie Mellon University where I have the good fortune of being co-advised by Professors Henny Admoni and Reid Simmons.

Prior to becoming a PhD student, I was a Master's student in the RI. During that time, I was co-advised by Professors Henny Admoni and Oliver Kroemer.

Before enrolling at CMU, I worked on the navigation and planning software for MoonRanger under Professors Red Whittaker and David Wettergreen, and I also worked on the CubeRover project's Systems Engineering team. Before robotics, I taught 9th grade Humanities and coached boys' lacrosse and ice hockey at Culver Academies. I earned a dual degree in English and Economics from the University of Virginia.

Email  /  CV  /  Google Scholar

profile photo

Research

These days, I'm especially interested in using machine learning and robotic manipulation to improve a robot's efficacy as an individual's long-term personal assistant. In particular, I aim to:

  1. make interactions with a robot as intuitive as possible for people,
  2. use human-provided feedback to improve a robot’s understandings of—and ability to interact with—its human partners and its environment, and
  3. validate our proposed methods through real-world user studies.

INQUIRE: INteractive Querying for User-aware Informative REasoning.
Tesca Fitzgerald, Pallavi Koppol, Patrick Callaghan, Russell Q. Wong, Reid Simmons, Oliver Kroemer, Henny Admoni.
Conference on Robot Learning, 2022
github  /  paper

We studied how an agent might actively select from multiple forms of human-provided feedback to learn the weights of a human's reward function.

Image and Lidar Dataset of the West Desert Sinkhole: An Analog for Steep-walled Planetary Pits.
Jordan Ford, Patrick Callaghan, Uland Wong, Heather Jones, Chuck Whittaker, William "Red" Whittaker.
International Planetary Caves Conference, 2020
paper

We went to the desert and collected data to model the moon's massive pits (which might one day serve as outposts for astronauts).


Thanks to Jon Barron for making his website's source code available to the rest of us!