Ankit J. Shah

PhD Candidate, MIT

Interactive Robotics Group

Computer Science and Artificial Intelligence Laboratory

Massachusetts Institute of Technology

Biography

I am a PhD candidate in the Interactive Robotics Group at MIT advised by Prof. Julie Shah. My research vision involves developing computational models and algorithms that allow domain-experts to directly train robots just as they would train a human apprentice. Leveraging human domain-experts’ strengths to enable robots to teach and learn effectively will be the defining theme of my long-term research. I believe that enabling domain-experts to directly train autonomous systems to perform complex tasks while enabling transparency in the system’s policy will be crucial enabler for widespread deployment of autonomous systems.

I am on the market for tenure-track faculty and industry research positions. You can find my research statement and the teaching philosophy statement here. Please reach out if you are hiring.

Interests

Machine Learning and AI for Robotics
Human-Machine Interaction
Ease of Robot Programming

Education

PhD in Aeronautics and Astronautics, 2021 (expected)

Massachusetts Institute of Technology
S.M. in Aeronautics and Astronautics, 2016

Massachusetts Institute of Technology
B.Tech. in Aerospace Engineering, 2013

Indian Institute of Technology - Bombay

Updates

March 2020: Our research on ‘Planning with Uncertain Specifications (PUnS)’ was featured on the MIT home page. Check out the article here.
Feb 2020: Our paper on ‘Planning with Uncertain Specifications (PUnS)’ has been accepted for publication in IEEE Robotics and Automation Letters (RA-L). It will also be presented at ICRA 2020.
Dec 2019: My abstract on ‘Interactive Training for Temporal Tasks’ has been accepted at the 2020 HRI Pioneers Workshop.
June 2019: I will be presenting our work on PUnS at RSS 2019 at the Workshop on Combining Learning and Reasoning (RSS-CLeAR) and at the workshop on Robust Autonomy
May 2019: I will be presenting my work on leveraging temporal logics for learning from demonstrations at University of Washington.
March 2019: I will be presenting my work on Bayesian specification inference at CU Boulder on March 21st.
December 2018: I will be presenting our paper at NeurIPS 2018 on December 4th.
October 2018: I will be presenting a talk on Bayesian specification inference at Brown University robotics on October 12th.
September 2018: The paper titled ‘Bayesian inference of temporal task specifications’ has been accepted for a poster presentation at NeurIPS 2018 in Montreal, Canada.

Recent Publications

All Publications

Ankit Shah, Julie Shah (2020). Interactive Robot Training for Temporal Tasks. HRI Pioneers, Companion of the 2020 ACM/IEEE International Conference on Human-Robot Interaction.

PDF

Ankit Shah, Samir Wadhwania, Julie Shah (2020). Interactive Robot Training for Non-Markov Tasks. arXiv:2003.02232 (under review).

PDF Project

Serena Booth, Yilun Zhou, Ankit Shah, Julie Shah (2020). Bayes-Probe: Distribution-Guided Sampling for Prediction Level Sets. arXiv:2002.10248 (under review).

PDF

Ankit Shah, Shen Li, Julie Shah (2020). Planning With Uncertain Specifications (PUnS). IEEE Robotics and Automation Letters (To be presented at ICRA 2020).

PDF Project Video

Serena Booth*, Ankit Shah*, Yilun Zhou*, Julie Shah (2020). Sampling Prediction-Matching Examples in Neural Networks: a Probabilistic Programming Approach. AAAI Workshop on Statistical Relational AI.

PDF

Patrick Craven, Kevin Oden, Kevin Landers, Ankit Shah, Julie Shah (2019). Man-Machine Interoperation in Training for Large Force Exercise Air Missions. Interservice/Industry Training, Simulation and Education Conference.

PDF

Ankit Shah, Julie Shah (2019). Notions of Safety with Uncertainty in Specifications. Robotics: Science and Systems, Workshop on Robust Autonomy – safe robot learning and control in uncertain real-world environments (Poster).

Project Poster

Ankit Shah, Julie Shah (2019). Planning with Uncertain Specifications. Robotics: Science and Systems, Workshop on Combining Learning and Reasoning – Towards Human-Level Robot Intelligence.

PDF Project Poster