Students and faculty affiliated with the THINC lab conduct cutting-edge research in AI and Robotics. We rigorously investigate various problems of interest in contexts such as multiagent systems, reinforcement learning, mobile robots, and the semantic web. We have published more than a hundred papers in top-tier research conferences such as AAAI, IJCAI, AAMAS, IROS, ICRA, and journals such as JAIR and JAAMAS. Our research is often multi-disciplinary and we collaborate with renowned researchers including psychologists, education specialists and biologists in the US and beyond. Our research has been funded by grants from NSF, multiple DoD agencies, NIH, and the industry.
Prashant Doshi
Dr. Prashant Doshi

Director, THINC Lab

About Us

The THINC Lab is housed in the Computer Science department of the University of Georiga. It was founded in July 2010 by Dr. Prashant Doshi who is a professor of computer science. It is located on the fifth floor of Boyd Graduate Studies Research Center in the heart of the south campus of UGA. Typically, about 10 graduate students and a few undergraduate students conduct their thesis, dissertation, or independent studies in the lab. Members of the lab are highly motivated and are encouraged to strive for excellence in research while being an integral part of a collegial setting. You are welcome to browse the lab’s webpages for additional information about our research.

Recent Publications

[2019] Saurabh Arora, Prashant Doshi, and Bikramjit Banerjee, Online Inverse Reinforcement Learning Under Occlusion, in AAMAS 2019. [paper] [supplement]

[2019] Vinamra Jain, Prashant Doshi, and Bikramjit Banerjee, Model-Free IRL using Maximum Likelihood Estimation, in AAAI 2019. [paper]

[2018] Tomoki Nishi, Prashant Doshi, and Danil Prokhorov, Merging in Congested Freeway Traffic using Multipolicy Decision Making and Passive Actor-Critic Learning", accepted in IEEE Transactions for Intelligent Vehicles, 2018.

[2018] Abdullah Rashwan, Agastya Kalra, Pascal Poupart, Prashant Doshi, and George Trimponias, Online Structure Learning for Feed-Forward and Recurrent Sum-Product Networks, in NIPS, 2018. [paper] [video]

[2018] Saurabh Arora and Prashant Doshi, A Survey of Inverse Reinforcement Learning: Challenges, Methods and Progress, arXiv:1806.06877v1

[2018] Maulesh Trivedi and Prashant Doshi, Inverse Learning of Robot Behavior for Collaborative Planning, in IROS, 2018. [paper] [video]

[2018] Kenneth Bogert and Prashant Doshi, Multi-Robot Inverse Reinforcement Learning under Occlusion with Estimation of State Transitions, in AI Journal, 2018. [article]

[2018] Roi Ceren, Prashant Doshi and Keyang He, Reinforcement Learning for Heterogeneous Teams with PALO Bounds, arXiv:1805.09267

[2018] Saurabh Arora, Prashant Doshi and Bikramjit Banerjee, A Framework and Method for Online Inverse Reinforcement Learning, arXiv:1805.07871

[See more..]

Follow Us

Latest News

3000 citations

[03/04/19] Papers published by THINC Lab researchers cross the 3,000 citations mark.

[10/11/18] THINC Lab welcomes Metonymize Inc. as an industrial collaborator. The funded collaboration will use SPNs as a generative and explainable model for moderately sized data.

NSF Grant

[08/20/18] New 3-year NSF NRI grant for inverse learning of robot behavior for collaborative tasks. Prof. Doshi (PI) will collaborate with Prof. Yi Hong (CS-UGA) and Prof. Ken Bogert at UNC-Asheville on this grant.


[07/15/18] New 3-year ARO grant on modeling cyberdeception and investigating a new framework for asymmetric interactions between attackers and defenders. Prof. Doshi (PI) will collaborate with Prof. Kyu Lee (CS-UGA) on this grant.

NSF Grant

[07/01/18] New 3-year NSF NRI grant on investigating sum-product networks for tractable data-driven decision making. NSF reviewers viewed the proposal as transformative. Prof. Doshi (PI) will collaborate with Prof. Pascal Poupart (UWaterloo, Canada) on this grant.

[06/06/18] Prof. Doshi gave an invited talk on I-POMDPs and its application toward modeling human behavior in strategic tasks at the Cross-Modal Learning Center of UKE, Hamburg, Germany.

[See more..]