Computing That Serves

Deploying Robots to Work In Concert with People – Human-Inspired Computational Techniques


Thursday, February 28, 2013 - 11:00am


Julie Shah

Assistant Professor, Department of Aeronautics and Astronautics, MIT


Mike Goodrich

Colloquium is in 1170 TMCB.

Robots are increasingly
introduced to work in concert with people in high-intensity domains, such as manufacturing,
military field operations, and disaster response. To harness the relative
strengths of humans and robots, we must develop robots that seamlessly
integrate with human group dynamics. Although there are
numerous studies on human teamwork and coordination in high-intensity domains, very
little prior work exists on applying these models to human-robot interaction.
In this talk, I describe ongoing work aimed at translating qualitative insights
from human factors engineering into quantitative, predictive models that
improve human-robot teamwork. I discuss two key challenges: learning team fluency
through experience and practice of repetitive tasks (e.g. on a manufacturing
assembly line), and pre-planning for team coordination in highly unstructured
domains (e.g. disaster first-response).

For learning fluency in repetitive joint-action,
I show that we can quantitatively formulate a human team training strategy to
emulate the mutual adaptation that occurs naturally in human team training.
Large-scale human subject experiments show that this human-robot interactive
training method provides statistically significant improvements in human-robot
team performance, as compared to the prior state-of-the-art in interactive
reinforcement learning. For pre-planning in unstructured domains, I present a
statistical approach to learning patterns of strong and weak agreements in
natural human team planning dialog, and show that this method achieves greater
than 90% prediction accuracy. I will also preview the next step in this
research effort: comprehensively tracking and summarizing plan formation (what the
team has agreed upon and temporal relationships in the plan) using Bayesian
modeling techniques. Encouraging results from these initial studies support the
hypothesis that human-inspired methods can reduce the burden of programming and
deploying autonomous systems to work in concert with people.


Julie Shah
is an Assistant Professor in the Department of Aeronautics and Astronautics and leads the Interactive Robotics
Group of the Computer Science and Artificial Intelligence Laboratory. Shah
received her SB (2004) and SM (2006) from the Department of Aeronautics and
Astronautics at MIT, and her PhD (2010) in Autonomous Systems from MIT. Before
joining the faculty, she worked at Boeing Research and Technology on robotics
applications for aerospace manufacturing. 
She has developed innovative methods for enabling fluid human-robot
teamwork in high-intensity domains, ranging from manufacturing to surgery to
space exploration. Her group draws on expertise in artificial intelligence,
human factors, and systems engineering to develop interactive robots that work
in concert with people. Ongoing projects in manufacturing include
collaborations with Boeing Research and Technology, ABB, and BMW, to develop
methods for fluidly coordinating human and robotic work on the factory floor.