Rajat Talak
I am a Research Scientist in the Department of Aeronautics and Astronautics at the Massachusetts Institute of Technology. I work in SPARKlab with Prof. Luca Carlone.
My big-picture research interests are in spatial robot perception, safe perception, and networked autonomy. My technical skills and expertise are in robot and computer vision, networks and graphs, optimization theory and algorithms, estimation theory and machine learning, geometric deep learning, stochastic control, differentiable optimization, uncertainty quantification, and domain adaptation.
I was a Postdoctoral Associate at MIT, from 2020-2022. In 2020, I received a Ph.D. in the field of Networked Autonomy, from the Department of Aeronautics and Astronautics at MIT. During this time, I worked with Prof. Eytan Modiano and Prof. Sertac Karaman in the Laboratory of Information and Decision Systems (LIDS). My work received ACM MobiHoc 2018 Best Paper Award.
News
15 Nov 2023: Guest Lectures at VNAV. You can access the course at MIT OCW [Link].
13 Nov 2023: Serving as TPC member for the Age and Semantics of Information (ASoI) Workshop at INFOCOM 2024.
04 Oct 2023: Co-Chairing a Session at IROS 2023.
04 Oct 2023: Paper Talk "Certifiable 3D Object Pose Estimation: Foundations, Learning Models, and Self-Training." [arXiv][IEEEXplore] at IROS 2023.
10 July 2023: Invited Talk: "Spatial Perception for Robotics: Representation, Structure, and Real-Time Systems" at RSS 2023 Workshop on Robot Scene Representations. [Workshop Link][YouTube].
10 July 2023: Safe Autonomy Workshop at RSS 2023. [Workshop Link].
17 May 2023: Pre-print release: "Foundations of Spatial Perception for Robotics: Hierarchical Representations and Real-time Systems." [arXiv][YouTube][GitHub]
24 Apr 2023: Accepted at IEEE Trans. Robotics: "Certifiable 3D Object Pose Estimation: Foundations, Learning Models, and Self-Training." [arXiv].
22 Apr 2023: Accepted at RSS 2023: "A Correct-and-Certify Approach to Self-Supervise Object Pose Estimators via Ensemble Self-Training." [arXiv].
19 Apr 2023: Organizing Workshop on Safe Autonomy at RSS 2023. [Workshop link].
27 Feb 2023: Accepted at CVPR 2023: "PyPose: A Library for Robot Learning with Physics-based Optimization." [arXiv][webpage][GitHub].
25 Feb 2023: Invited talk: "Information Freshness for Networked Autonomy" at the 29th National Conference on Communication, IIT Guwahati. [YouTube].
06 Feb 2023: Talk: "Certifiable 3D Object Pose Estimation: Foundations, Learning Models, and Self-Training" at Stanford Intelligent Systems Lab.
25 Jan 2023: Pre-print release: "Certifiable 3D Object Pose Estimation: Foundations, Learning Models, and Self-Training." [arXiv].
12 Dec 2022: Serving as TPC member for the 6th Age of Information Workshop at INFOCOM 2023. [Workshop Page].
11 Nov 2022: Invited talk: "A Vision for Certifiable Perception: from Outlier-Robust Estimation to Self-Supervised Learning" at NeurIPS 2022. [Video][Workshop Link].
Contact
talak@mit.edu, rajatalak@gmail.com
MIT Office 31-219, 77 Massachusetts Avenue, Cambridge, MA 02139, USA