Pushkal Katara

SRM Institute of Science and Technology, Carnegie Mellon University

Papers

Total Citations

H-Index

About

Pushkal Katara is a robotics researcher whose work spans underwater robotics simulation, robot learning, and vision-based control systems. He made an early mark with URSim, an open-source 3D simulation framework for Unmanned Underwater Vehicles (UUVs) built on ROS and Unity3D, which has garnered 40 citations and become a valuable tool for researchers seeking accessible platforms to implement and test complex underwater robotic systems. Katara's research trajectory then expanded into the frontier of robot learning at scale, with his 2024 work Gen2Sim demonstrating how generative models can dramatically reduce human effort in building simulation environments and reward functions — enabling generalist robot manipulators to acquire diverse manipulation skills with far less manual intervention, accumulating 18 citations in a short time. His work on DeepMPCVS further illustrates his breadth, combining deep learning with model predictive control to improve visual servoing performance in unseen environments. Across these contributions, Katara consistently addresses bottlenecks in practical robotics — from accessible simulation to scalable training pipelines — positioning him as a thoughtful contributor to both the infrastructure and intelligence layers of modern autonomous systems.