Prof. Gregory J. Stein
RAIL Group, GMU

I am an Assistant Professor of Computer Science at George Mason University where I run the Robotic Anticipatory Intelligence & Learning (RAIL) Group. Our research, at the intersection of robotics and machine learning, is centered around developing representations that allow robots to better understand the impact of their actions, so that they may plan quickly and intelligently in a dynamic and uncertain world.

Read more on our research page or see our full publication list.

See also my curriculum vitae (CV).

Recent and Highlighted News

See all news.

Selected Publications

See also our full publication list.

@inproceedings{li2023modelexplore,
  author =     {Yimeng Li and Arnab Debnath and Gregory J. Stein and Jana
                Kosecka},
  title =      {Learning-augmented model-based planning for 
                visual exploration},
  booktitle =  {International Conference on Intelligent Robots 
                and Systems (IROS)},
  year =       {2023},
  note =       {in press},
}
@inproceedings{arnob2023lspgnn,
  title =      {Improving Reliable Navigation under Uncertainty via 
                Predictions Informed by Non-Local Information},
  author =     {Arnob, Raihan Islam and Stein, Gregory J.},
  booktitle =  {International Conference on Intelligent Robots 
                and Systems (IROS)},
  year =       {2023},
  note =       {in press},
}
@inproceedings{paudel2023selection,
  title =      {Data-Efficient Policy Selection for Navigation in 
                Partial Maps via Subgoal-Based Abstraction},
  author =     {Paudel, Abhishek and Stein, Gregory J.},
  booktitle =  {International Conference on Intelligent Robots 
                and Systems (IROS)},
  year =       {2023},
  note =       {in press},
}
@inproceedings{khanal2023guided,
  title =     {Guided Sampling-Based Motion Planning with Dynamics 
               in Unknown Environments},
  author =    {Khanal, Abhish and Bui, Hoang-Dung and Stein, Gregory J. and Plaku, Erion},
  booktitle = {International Conference on Automation Science 
               and Engineering (CASE)},
  year =      {2023},
  note =      {in press},
}
@inproceedings{dhakal2023anticipatory,
  title =      {Anticipatory Planning: Improving Long-Lived Planning 
                by Estimating Expected Cost of Future Tasks},
  author =     {Dhakal, Roshan and Talukder, Md Ridwan Hossain 
                and Stein, Gregory J.},
  booktitle =  {International Conference on Robotics and Automation (ICRA)},
  pages =      {11538--11545},
  year =       {2023},
}
@inproceedings{khanal2023mrlsp,
  title =      {Learning Augmented, Multi-Robot Long-Horizon Navigation in
                Partially Mapped Environments},
  author =     {Khanal, Abhish and Stein, Gregory J.},
  booktitle =  {International Conference on Robotics and Automation (ICRA)},
  pages =      {10167--10173},
  year =       {2023},
}
@inproceedings{stein2021xailsp,
  title =      {Generating High-Quality Explanations for Navigation 
                in Partially-Revealed Environments},
  author =     {Gregory J. Stein},
  booktitle =  {Advances in Neural Information Processing Systems (NeurIPS)},
  pages =      {17493--17506},
  year =       {2021},
  keywords =   {explainability; planning under uncertainty; 
                subgoal-based planning; interpretable-by-design},
}
@inproceedings{bradley2021potlp,
  title =     {Learning and Planning for Temporally Extended Tasks in Unknown
               Environments},
  author =    {Christopher Bradley and Adam Pacheck and Gregory J. Stein and
               Sebastian Castro and Hadas Kress-Gazit and Nicholas Roy},
  booktitle = {International Conference on Robotics and Automation (ICRA)},
  pages =     {4830--4836},
  year =      {2021},
}
@inproceedings{stein2018subgoal,
  author =    {Stein, Gregory J. and Bradley, Christopher and Roy, Nicholas},
  title =     {Learning over Subgoals for Efficient Navigation of Structured,
               Unknown Environments},
  booktitle = {Conference on Robot Learning (CoRL)},
  pages =     {213--222},
  year =      {2018},
}

Best Paper Finalist at CoRL 2018; Best Oral Presentation at CoRL 2018.

Blog Posts

  • When should PhD students meet with their advisor? A guide
    [Communication & Learning] Why have meetings with my advisor? This document gives an overview of how I think about when and how often my PhD students need to meet with me and what the substance of those meetings should be.
  • PaperOps: run experiments and add results to a PDF with a single Make command
    [Workflow & Process] A proof-of-concept for automated running of experiments and automatically including results and statistics in a LaTeX PDF via Make.
  • Improving Reliable Navigation under Uncertainty with Non-Local Information Informed Predictions
    [Research] We improve reliable, long-horizon, goal-directed navigation in partially-mapped environments by using non- locally available information to predict the goodness of temporally-extended actions that enter unseen space.
  • Data-Efficient Policy Selection for Navigation in Partial Maps
    [Research] We present a fast and reliable policy selection approach for navigation in partial maps that leverages information collected during deployment to introspect the behavior of alternative policies without deployment.
  • A Communication Exercise: randomize slide presenters
    [Communication & Learning] I recently asked my students to prep Pecha Kucha-style presentations—18 slides that auto-advance every 20 seconds—and then told them to present each other’s slides, a fun and fantastic (if chaotic) way to improve slide quality.
  • Seek out Opportunities to Mentor
    [Workflow & Process] Mentoring others can be incredibly valuable experience and so opportunities to mentor should be sought out.
  • My strategies for idea generation through writing
    [Workflow & Process] This guide aims to outline my strategies to encourage idea generation, essential for long-term research progress, through a habit of regular writing.
  • Write code and tests in tandem
    [Workflow & Process] Even research code benefits greatly from automated testing. My approach to getting started: if you need to write additional code to verify that some functionality is working, that additional code should be written as a test.
  • Generating high-quality explanations for navigation in partially-revealed environments
    [Research] We generate explanations of a robot agent’s behavior as it navigates through a partially-revealed environment, expressed in terms of changes to its predictions about what lies in unseen space. Blog post accompanying our NeurIPS 2021 paper.
  • Underpromise and overdeliver to your future self
    [Workflow & Process] I’ve started to think of expectation management as a part of self-care, and I try to think of myself as an other. As such, I try to underpromise and overdeliver to my future self.
  • Action items should have well-defined end conditions
    [Workflow & Process] I try to make sure that my tasks are both easy to start making progress towards and have clear completion criteria, essential criteria for making sure they get done.
  • Reviewing papers is incredibly valuable experience
    [Communication & Learning] Looking at the in-development work of others pulls back the curtain and reveals insight into the thought process of other researchers. Reviews are a largely-untapped pedagogical resource.
  • Accelerating team research with containers
    [Workflow & Process] All code in my research group is run exclusively inside Docker containers, helping us develop more quickly and share code with ease.
  • Talk figures are different from paper figures
    [Communication & Learning] One communication pitfall I often see is that many researchers will take figures from their papers and paste them into their slides. Here, I provide some tips for tailoring your figures to talks.
  • Machine Learning & Robotics: My (biased) 2019 State of the Field
    [Research] My thoughts on the past year of progress in Robotics and Machine Learning (2019).
  • DeepMind's AlphaZero and The Real World
    [Research] Using DeepMind’s AlphaZero AI to solve real problems will require a change in the way computers represent and think about the world. In this post, we discuss how abstract models of the world can be used for better AI decision making and discuss recent work of ours that proposes such a model for the task of navigation.



Gregory J Stein
Assistant Professor
Computer Science

gjstein@gmu.edu
home research
curriculum vitae (CV)