I am an Assistant Professor in the Departments of Electrical & Computer Engineering (ECE) and Mathematics and the Khoury College of Computer Sciences (by courtesy) at Northeastern University, where I lead the Robust Autonomy Lab.
I am broadly interested in the mathematical and algorithmic foundations of trustworthy autonomy. My research applies analytical and computational tools from nonlinear optimization, differential geometry and topology, abstract algebra, and probability and statistics to design principled, computationally efficient, and provably robust algorithms for machine learning, perception, and control.
A major focus of my research is the design of practical estimation and control algorithms for nonlinear systems that provide explicit performance guarantees in real-world operation. To that end, much of my recent work has explored the use of convex relaxation as a general strategy for recovering provably-good approximate solutions to hard computational problems in artificial intelligence. Check out our recent survey article if you’d like to know more!
ScD Computer Science, 2016
Massachusetts Institute of Technology
MA Mathematics, 2010
University of Texas at Austin
BS Mathematics, 2008
California Institute of Technology
Honorable Mention, King-Sun Fu Memorial Best Paper Award, 2021
IEEE Transactions on Robotics
Best Student Paper Award, 2020
Robotics: Science and Systems
RSS Pioneer, 2019
Robotics: Science and Systems
Best Paper Award, 2016
International Workshop on the Algorithmic Foundations of Robotics
[15-Jan-2024] Our research group has been awarded a new grant from the Charles Stark Draper Laboratory to support our work on decentralized collaborative multi-agent perception! Many thanks to Draper for their support :-)!
[28-Nov-2023] Our new paper applies Fourier analysis on compact groups to develop a computationally-efficient (convex) relaxation for a very general family of robust synchronization problems over compact groups, including (in particular) the fundamental problem of robust rotation averaging. Our approach affords several notable advantages versus prior RA methods: it can be used in conjunction with any smooth loss function, does not require initialization, and is implemented using only simple (and highly scalable) linear-algebraic computations and parallelizable optimizations over functions of individual rotational states. Congrats to lead authors Owen Howell and Haoen Huang for an outstanding piece of work :-)!
[28-Sep-2023] I presented some of our group’s recent work on certifiable estimation at UIUC’s Allerton Conference on Communication, Control, and Computing in the session on “Advances in Nonconvex Optimization”. Many thanks to Richard Zhang, and Salar Fattahi for organizing a fantastic program, and for the invitation to speak!
[19-Sep-2023] Our new paper describes a simple and computationally-efficient approach to directly optimizing the design of sensor packages for mobile robot navigation. Congrats to lead author Pushyami Kaveti for a very cool piece of work :-)!
[3-Sep-2023] Our research group has been awarded a new grant from MIT’s Lincoln Laboratory to support our development of computational tools for certifiable perception, planning, and control. Many thanks to MITLL for their support :-)!