Michael Everett

mfe@mit.edu
31-235C
Google Scholar
GitHub
YouTube Playlist

About

I am a Postdoctoral Associate at MIT Aero/Astro in the Aerospace Controls Lab. My research lies at the intersection of robotics, deep learning, and control theory. I received the PhD (2020), SM (2017), and SB (2015) degrees from MIT in Mechanical Engineering. My PhD work was advised by Prof. Jonathan How, Prof. John Leonard, and Prof. Alberto Rodriguez.

Selected Awards

  • Winner: Best Paper Award on Cognitive Robotics (IROS 2019)
  • Winner: Best Student Paper (IROS 2017)
  • Finalist: Best Paper Award on Cognitive Robotics (IROS 2017)
  • Finalist: Best Multi-Robot Systems Paper (ICRA 2017)
  • iCampus Student Prize Winner ($3,000) for ofcourse.mit.edu

Research Vision

My research aims to spark a new era of autonomy defined by robots that are resilient, dependable, and ready to support humans throughout the real world.

To get there, my goal is to revolutionize reliability in robot learning by constructing key pillars of resiliency:

  • Safety Guarantees of Neural Feedback Loops
    • learning safe-by-construction policies
    • interaction of learning and classical modules
  • Assured Performance of Learned Policies
    • principled sim-to-real transfer
    • addressing data starvation at human-autonomy interface
  • Super-Human Abilities
    • high speed, off-road autonomy

My PhD and postdoctoral research builds foundations of:

  • Autonomy in human environments
    • Socially aware navigation
    • Context-guided exploration
  • Robust deep learning
    • Uncertainty analysis of deep neural networks
    • Certifiably robust deep RL
Neural Feedback Loop (NFL)

Publications


Pre-prints

FASTER: Fast and Safe Trajectory Planner for Flights in Unknown Environments

Jesus Tordesillas Torres, Brett T. Lopez, Michael Everett, Jonathan P. How
IEEE Transactions on Robotics (TRO), in review
Paper


Refereed Papers

Efficient Reachability Analysis of Closed-Loop Systems with Neural Network Controllers

Michael Everett, Golnaz Habibi, Jonathan P. How
ICRA 2021 (accepted)
Paper     Code

Certifiable Robustness to Adversarial State Uncertainty in Deep Reinforcement Learning

Michael Everett*, Björn Lütjens*, Jonathan P. How
IEEE Transactions on Neural Networks and Learning Systems (TNNLS), 2021 (accepted)
IEEE Xplore     Arxiv

Collision Avoidance in Pedestrian-Rich Environments with Deep Reinforcement Learning

Michael Everett, Yu Fan Chen, Jonathan P. How
IEEE Access, Vol. 9, pp. 10357-10377, January 2021.
Special Section: "Real-Time Machine Learning Applications In Mobile Robotics"
IEEE Xplore     Arxiv     Code: [ Pre-Trained ROS Package , Training Environment , RL Training Code ]

Where to go next: Learning a Subgoal Recommendation Policy for Navigation in Dynamic Environments

Bruno Brito, Michael Everett, Jonathan P. How, Javier Alonso-Mora
IEEE Robotics & Automation Letters (RA-L) 2021 (accepted)
ICRA 2021 (accepted)
Paper     Code (coming soon...)

Robustness Analysis of Neural Networks via Efficient Partitioning with Applications in Control Systems

Michael Everett, Golnaz Habibi, Jonathan P. How
IEEE Control Systems Letters 2020 (accepted)
ACC 2021 Invited Session on Learning, Optimization, and Control for Safety-critical Systems (accepted)
IEEE Xplore     Arxiv     Code

Multi-agent Motion Planning for Dense and Dynamic Environments via Deep Reinforcement Learning

Samaneh Hoseini, Hugh H.T. Liu, Michael Everett, Anton de Ruiter, Jonathan P. How
IEEE Robotics & Automation Letters (RA-L)
Vol. 5, No. 2, pp. 3221-3226, April 2020
IEEE Xplore     Arxiv

Planning Beyond The Sensing Horizon Using a Learned Context

Michael Everett, Justin Miller, Jonathan P. How
IROS 2019
Winner: Best Paper Award on Cognitive Robotics
Paper     Code     Video

Certified Adversarial Robustness in Deep Reinforcement Learning

Björn Lütjens, Michael Everett, Jonathan P. How
Conference on Robot Learning (CoRL) 2019
Paper     Talk

R-MADDPG for Partially Observable Environments and Limited Communication

Rose E Wang, Michael Everett, Jonathan P How
Reinforcement Learning for Real Life (RL4RealLife) Workshop in ICML 2019
Paper     Code

Safe Reinforcement Learning with Model Uncertainty Estimates

Björn Lütjens, Michael Everett, Jonathan P. How
ICRA 2019
Selected for Oral Presentation at IROS 2018 Workshop on Machine Learning in Motion Planning
Paper

Motion Planning Among Dynamic, Decision-Making Agents with Deep Reinforcement Learning

Michael Everett, Yu Fan Chen, Jonathan P. How
IROS 2018
Paper     Code     Video

Socially Aware Motion Planning With Deep Reinforcement Learning

Yu Fan Chen, Michael Everett, Miao Liu, Jonathan P. How
IROS 2017
Winner: Best Student Paper Award
Finalist: Best Paper Award on Cognitive Robotics
Paper     Video

Decentralized Non-Communicating Multiagent Collision Avoidance With Deep Reinforcement Learning

Yu Fan Chen, Miao Liu, Michael Everett, Jonathan P. How
ICRA 2017
Finalist: Best Multi-Robot Systems Paper
Paper     Video

Scalable Accelerated Decentralized Multi-Robot Policy Search in Continuous Observation Spaces

Shayegan Omidshafiei, Christopher Amato, Miao Liu, Michael Everett, Jonathan P How, and John Vian
ICRA 2017
Paper    

Semantic-level Decentralized Multi-Robot Decision-Making using Probabilistic Macro-Observations

Shayegan Omidshafiei, Shih-Yuan Liu, Michael Everett, Brett T. Lopez, Christopher Amato, Miao Liu, Jonathan P. How, John Vian
ICRA 2017
Paper     Video

Seeing around corners with a mobile phone? Synthetic aperture audio imaging

Hisham Bedri, Micha Feigin, Michael Everett, Ivan Filho, Gregory L. Charvat, Ramesh Raskar
ACM SIGGRAPH 2014 Posters
Paper

Theses

Algorithms for Robust Autonomous Navigation in Human Environments

Michael Everett
PhD Thesis, MIT
Paper     Defense Talk

Robot Designed for Socially Acceptable Navigation

Michael Everett
SM Thesis, MIT
Paper

Open-Source Software



CADRL ROS package


Star Fork

Collision Avoidance Training Environment (Gym)


Star Fork

RL Training for Collision Avoidance (GA3C-CADRL)


Star Fork


Loomo Android-ROS


Star Fork

Deep Cost-to-Go


Star Fork

Trajectory Overlays


Star Fork


Neural Network Robustness Analysis


Star Fork





Featured Press