Pranav Malpure

Hello! I'm Pranav, a Graduate Student Researcher at UC San Diego, where I am pursuing my MS in ECE, specializing in Intelligent Systems, Robotics, and Control (ISRC). I completed my undergraduate from the Indian Institute of Technology Bombay, receiving an Honors degree in Aerospace Engineering and a Minor degree in Systems and Controls Engineering.

My major work/research focus lies in reinforcement learning in robotics, exploring different ideas in manipulation, currently working on its integration with language for better performances. I've also exploreed multi-robot navigation, sensor-fusion based SLAM, control systems, and embedded electronics. I am currently a researcher at the Existential Robotics Laboratory, working under Prof. Nikolay Atanasov. My undergraduate thesis was on the application of Deep RL for control of a robotic arm to achieve inverse kinematics, and was advised by Prof. Mayank Baranwal.

I am curious to see how robots will shape our everyday lives in the future, and want to be a part of the community that makes that happen. Safe, friendly and useful robots in our homes for the win!

Research Interests: Reinforcement Learning, Dexterous manipulation, VLMs, Visual Learning

Email  /  GitHub  /  Resume  /  LinkedIn

M.S. in ECE(Robotics)

Sep '24 - Dec '25

Robotics Intern

May '23 - Jul '23

B.Tech(Honors) in Aerospace

Nov '20 - May '24

Graduate Research: Reward Engineering of allegro hand for cube grab

Reward engineered an allegro hand attached to a xarm robot, comprising of 21 DoF, to grab a cube based on RGBD data as observation. This involved breaking the rewards into stages, each accomplishing certain segments of the entire task.

Bachelor's Thesis: Deep Reinforcement Learning for control of robotic arm for Inverse Kinematics

Implemented a model-free reinforcement learning approach to train control policies for a robotic arm to navigate to any point in its workspace. Developed a reward structure in dm_control for a kinova Jaco arm.

Multi-Robot navigation in Cyclic Graphs

In this proprietary work, I found the shortest time required by the MR DFS algorithm to explore unknown cyclic graphs by placing multiple robots at optimal nodes in the graph.

Autonomous underwater navigation

Programmed RexROV2 to autonomously navigate underwater using curvature velocity method to avoid obstacles, detected using the onboard sonar sensors. Simulated this in an underwater world using Gazebo in ROS

The Humanoid Project

Led the revival and growth of the Humanoid team at IIT Bombay, securing its first funding, recruiting two successive student batches, and establishing a long-term technical and organizational roadmap. Spearheaded cross-subsystem coordination and mentored juniors hands-on, providing strategic and technical direction to ensure sustained development and execution.

Optimizing recalculation of Visibility Graphs

Devised a novel technique to optimize the addition of new polygons in existing visibility graphs for drones. Reduced the recalculation time by around 92%.

credits to