Shivam Vats

I am a Postdoctoral Research Associate at the Intelligent Robot Lab led by Prof. George Konidaris in the Department of Computer Science, Brown University. I earned my PhD from the Robotics Institute at CMU where I was co-advised by Prof. Maxim Likhachev and Prof. Oliver Kroemer. I develop algorithms for robots to learn new skills from minimal supervision, search to creatively solve complex problems and continually improve with self-practice. This requires strategic robot learning to decide what to learn—focusing on the right policies and models—and how to learn—combining strategies like imitation learning, reinforcement learning and search to solve tasks resource-efficiently. This ASME article provides a good overview of my work on collaborative robots that plan what to learn.
Previously, I studied maths and computing at Indian Institute of Technology (IIT) Kharagpur where I worked with Prof. P.P. Chakrabarti and Prof. Bibhas Adhikari on heuristic search. I was a core developer of SymPy and an active contributor to SymEngine for many years. I also regularly participated in the Intelligent Ground Vehicle Competition as a member of the Autonomous Ground Vehicle Group.
I am looking for new collaborations and am always happy to chat about robotics and research. Do reach out if you are interested in working with me, especially if you are a student at Brown or MIT. Here are some of my ongoing projects:
- learning and planning with particle-based world models
- learning composable manipulation skills from videos
- multi-task interactive learning from human feedback
news
Jun 15, 2025 | RecoveryChaining: Learning Local Recovery Policies for Robust Manipulation has been accepted to IROS 2025. |
---|---|
May 25, 2025 | Our paper on Optimal Interactive Learning on the Job via Facility Location Planning has been accepted to RSS 2025. |
Mar 1, 2025 | Our paper on Multi-Robot Motion Planning with Diffusion Models was selected for Spotlight presentation at ICLR. |
Jul 28, 2024 | Invited talks at IISc and IIT Madras. |
May 23, 2024 | I defended my thesis on Plan to Learn: Active Robot Learning by Planning (video) |