Ram Natarajan

I am a Ph.D. candidate in The Robotics Institute at Carnegie Mellon University co-advised by Maxim Likhachev and Howie Choset.

I study and develop algorithms that explore the deeper connections between discrete search and continuous optimization for motion planning and long-horizon reasoning in robotics.

Prior to my Ph.D., I obtained my Master's in Robotics from Worcester Polytechnic Institute (WPI) where I was advised by Michael Gennert.

Email  /  CV  /  Scholar  /  GitHub  /  LinkedIn

A scalable and efficient way to plan on Graphs of Convex Sets (GCS) with stronger theoretical properties. We plan on GCS using a previously developed hybrid search-optimization framework called INSAT.

Inspired by the recent work on edge-based parallel graph search, PINSAT introduces systematic parallelization in INSAT to achieve lower planning times and higher success rates, while maintaining significantly lower costs.

An extension of INSAT framework for planning through contact with several algorithmic optimizations to combat high-dimensional search space and combinatorial contact modes.

Real-time projectile interception using constant-time kinodynamic motion planning. The system is made of an industrial manipulator holding a shield and a stereo camera for live perception.

A framework for simulataneous and systematic exploration of contact modes and dynamically feasible contact-rich trajectory generation.

Introduces a motion planning framework called INterleaved Search And Trajectory optimization (INSAT) for global, long-horizon reasoning through nonconvex spaces.

Extends Multi-Heurisitic A* (MHA*) to an anytime version. A-MHA* will quickly find a solution and continues improving it until timeout.

An efficient way to reuse the ordering computed during variable elimination when combining multiple factorized graphs. We demonstrate our method using loop-closure in multi-robot mapping.

An efficient way to reuse the ordering computed during variable elimination when combining multiple factorized graphs. We demonstrate our method using loop-closure in multi-robot mapping.

Website made based on Jon Barron's template.