1467787395758.jpg

Biography

I am a fifth year Ph.D. candidate studying computer science advised by Dr. Shlomo Zilberstein at the University of Massachusetts Amherst specializing in artificial intelligence applied to autonomous robots. My current research focuses is on designing large-scale automated planning and learning models for these partially observable stochastic systems, with applications to mobile humanoid service robots and autonomous vehicles. Specifically, I work with the Laboratory for Perceptual Robotics' humanoid uBots and Nissan Research Center's autonomous vehicles. Prior to this work, I spent a few years performing R&D on multiagent learning algorithms applied to real-world robotic scenarios at the Applied Research Laboratory. I earned bachelor's and master's degrees in computer science, engineering, and mathematics from the University of Massachusetts Amherst and The Pennsylvania State University. Please see my CV for more details.

Research Interests

My research interests are contained within the domains of automated planning, autonomous robotics, and machine (reinforcement) learning, in both single- and multi-agent scenarios. In particular, I am interested in the mathematically principled design of specialized MDPs, POMDPs, and Dec-POMDPs to solve complex real-world problems with hierarchical and multiple objective structures. Additionally, I develop highly parallelizable algorithms to solve them, both offline and online, for use in applied robotic settings such as high degree of freedom path/motion planning and real-time decision-making with integrated learning.

News

  • April 2018: Our IJCAI 2018 paper on meta-reasoning for determining the optimal stopping point in anytime algorithms. It has another video of the 'epic' library for path planning from our IROS 2016 paper!
  • January 2018: I am the primary inventor on another five international patent applications for autonomous vehicles!
  • November 2017: Our paper on integrating cooperation and competition (Dec-POMDP and POSG) was accepted to AAAI 2018! It has a multi-robot video using a hybrid Dec-POMDP and POSG with stochastic FSC policies!
  • July 2017: A paper on a very simple belief point compression technique called a σ-approximation was accepted to IROS 2017. It has an accompanying video of a robot using a POMDP!
  • May 2017: I am going to Nissan Research Center in Silicon Valley again to continue to develop decision-making algorithms for their autonomous vehicles!
  • April 2017: Our conference paper on MODIA, which is a method for online scalable decision-making with applications to autonomous vehicles, was accepted to IJCAI 2017!
  • April 2017: A paper on a simple belief point compression technique was accepted the ICAPS 2017 Planning and Robotics (PlanRob) workshop.
  • February 2017: I am the primary inventor on four international patent applications for autonomous vehicles!
  • November 2016: I am a co-organizer for the Multi-Objective Decision Making (MODeM) workshop at AAMAS 2017!
  • November 2016: Our AAAI 2017 paper on a short-sighted labeling algorithm for quickly solving SSPs entitled FLARES was accepted.
  • July 2016: News articles regarding our IJCAI semi-autonomous systems paper has appeared in Phys.org, EurekaAlert!, Science Codex, Science Daily, UMass News, and Daily Hampshire Gazette.
  • July 2016: Our IROS 2016 paper on robot path planning using harmonic functions and GPUs was accepted. This includes the 'epic' library for path planning in ROS.
  • May 2016: This summer I am at the Nissan Research Center in Silicon Valley working on their autonomous vehicles!
  • April 2016: Our IJCAI 2016 paper on semi-autonomous systems (namely vehicles) was accepted! Also, our IJCAI Goal Reasoning Workshop paper on a new SSP algorithm was accepted.
  • January 2016: Our AAMAS 2016 paper was accepted as an extended abstract.
  • December 2015: I have passed my qualifying exam with distinction! Now I am a Ph.D. candidate!
  • November 2015: Our AAAI 2016 paper on proactive learning POMDPs has been accepted.
  • August 2015: Both a GPU-based POMDP algorithm paper and a relaxed LMDP paper have been accepted to the SDMIA as part of the AAAI Fall Symposium 2015 Series. The POMDP short paper presents our 'nova' library for planning.
  • April 2015: Our IJCAI 2015 paper on Lexicographic POMDPs has been accepted.
  • February 2015: News articles published on our semi-autonomous driving work: NSF.gov, Phys.org, and Slashdot.org.
  • November 2014: Our AAAI 2015 paper on Lexicographic MDPs has been accepted.
  • May 2014: Two multiagent systems papers have been accepted to the MIPC workshop as part of AAAI 2014.
  • March 2014: I have received the Outstanding Teaching Assistant Award!