[Ml-stat-talks] Fwd: [ORFE-Seminars] TODAY: Optimization Seminar: Anirudha Majumdar, Princeton University, MAE: 4:30pm Sherrerd Hall 101

Barbara Engelhardt bee at princeton.edu
Thu Nov 16 11:04:59 EST 2017


Talk of interest.

*-----   TODAY:  Princeton Optimization Seminar   -----*



*Date:*          *Thursday, November 16 *



*Time:          *4:30PM



*Location:*   Sherrerd Hall 101



*Speaker:    Anirudha Majumdar, Princeton University*


Title:    *Optimization-based Techniques for Controlling Agile Robots with
Safety Guarantees*

Abstract:  The goal of my research is to develop algorithmic techniques for
controlling highly agile robotic systems such as unmanned aerial vehicles
while guaranteeing that they operate in a safe and reliable manner.

In this talk, I will describe our work on convex optimization-based
algorithms for the synthesis of feedback controllers that come with
associated formal guarantees on the stability of the robot and show how
these controllers and certificates of stability can be used for robust
planning in environments previously unseen by the system. In order to make
these results possible, we leverage powerful computational tools such as
sum of squares (SOS) programming and semidefinite programming, along with
approaches from nonlinear control theory. I will describe this work in the
context of high-speed unmanned aerial vehicle flight through cluttered
environments. The resulting hardware demonstrations on a small fixed-wing
airplane constitute one of the first examples of guaranteed safe and robust
control for robotic systems with complex nonlinear dynamics that need to
plan in realtime in environments with complex geometric constraints.

I will also present our recent work on ensuring safety in applications
where robots and humans interact. In particular, I will describe our work
on Risk-sensitive Inverse Reinforcement Learning (IRL) for modeling,
inferring, and predicting the behavior of humans operating in a robot’s
environment. In contrast to prior work on IRL that assumes that humans are
risk neutral, our approach is able to infer humans’ risk preferences from
observations of their actions. This allows us to better predict and imitate
the human decision making process. We believe that such an approach is an
important step towards being able to ensure safety in domains where robots
and humans interact.

Bio:  Anirudha Majumdar is an Assistant Professor in the Mechanical and
Aerospace Engineering (MAE) department at Princeton University. He received
a Ph.D. in Electrical Engineering and Computer Science from the
Massachusetts Institute of Technology in 2016, and a B.S.E. in Mechanical
Engineering and Mathematics from the University of Pennsylvania in 2011.
Subsequently, he was a postdoctoral scholar at Stanford University from
2016 to 2017 at the Autonomous Systems Lab in the Aeronautics and
Astronautics department. His research has been recognized with the Best
Conference Paper Award at the International Conference on Robotics and
Automation (ICRA) 2013, and the Siebel Foundation Scholarship.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cs.princeton.edu/pipermail/ml-stat-talks/attachments/20171116/71768edf/attachment.html>


More information about the Ml-stat-talks mailing list