About me
News:
-
July 2025: 🎉 Our work “Long Range Navigator (LRN)” was accepted to CoRL 2025! (Paper, Video)
-
July 2025: 🎉 Our work “WheeledLab” was accepted to CoRL 2025! (Paper, Website)
-
May 2025: 📝 Invited to review for CoRL
-
September 2024: 📝 Invited to review for RA-L
-
April 2024: 🎉 Our work “HOUND” was accepted in R: SS 2024! (Paper, Website, Video)
-
October 2023: 📝 Invited to review for ICRA
-
July 2023: 🎉 Our work “PuSHR” was accepted in IROS 2023! (Paper, Github)
-
September 2022: 🎓 Started my Ph.D. at the University of Washington!
I am a CS Ph.D. student at the University of Washington Seattle, part of Dr. Siddhartha Srinivasa's Personal Robotics Lab. I’m particularly interested in solving problems that enable/improve the real-world deployment of robots. This means that I tend to think about problems at multiple levels, ranging from system-level solutions down to individual subcomponents. For instance, in off-road autonomy research, the cost of the system-as-a-whole is a bottleneck. This cost can be reduced through software-based safety systems, and the H.O.U.N.D. (RSS ‘24, Paper, Website, Video) project demonstrates this with over 50 kilometers of real-world off-road driving. This project has enabled some of our other research projects, such as the WheeledLab (CoRL 25, Paper, Website), and components from this project have been utilized in our Long-Range Navigation (CoRL 25, Paper, Video) project.
However, sometimes the bottleneck is the compute constraint at a specific level; planning in continuous spaces is incredibly demanding due to the rapid growth of the search tree, and the process is inherently sequential. Without fast planning, the robot can struggle to navigate in real environments. In search-based approaches, the general approach to addressing the rapid growth of states is to discretize the state space and prune the tree; however, if you get this discretization wrong, you can either end up with no solution or a very computationally expensive one (or anything in between). Incremental Generalized Hybrid A* (in submission, Paper, Video) is a framework for efficient anytime planning in continuous spaces that addresses this problem.
Before starting my Ph.D., I worked at the Personal Robotics Lab with Dr. Christoforos Mavrogiannis and Dr. Siddhartha Srinivasa on a multi-agent non-prehensile manipulation project called PuSHR (IROS ‘23, Paper), which was a systems-level solution to coordinated rearrangement of objects using robots with nonholonomic constraints. I’ve also worked at the Indian Institute of Technology on their self-driving car project with Dr. Sunil Jha, where I worked on the state estimation and perception systems, and on V2X communication-based ADAS under Dr. Aakanksha Chowdhery.
I have also TA'd for the CSE478: Autonomous Robotics at UW in Spring '23, '24, '25.
If you're interested in seeing the projects I've made (pre-Ph.D.) from the ground up, peruse my project blogs here or on the top right corner by clicking on "Project-stories". There is a quick preview of those projects at the bottom of this webpage.
