Srinath Mahankali

I am currently a Research Engineer at Eka Robotics, where I work on sim-to-real reinforcement learning for dexterous robotic manipulation.

Previously, I finished my undergraduate at MIT, where I spent three years as a researcher in CSAIL advised by Pulkit Agrawal, developing energy-efficient controllers for quadruped locomotion and improving exploration in deep RL. Prior to that, I worked on inverse problems with Yunan Yang and studied the linear separability of one-layer neural networks with Promit Ghosal.

Updates

Research

We improve exploration in both discrete and continuous control domains by optimizing random reward functions parameterized by a sampled latent vector.

We train energy-efficient policies for quadruped locomotion tasks while improving task performance through constrained reinforcement learning.

Randomly generated rewards can explain a significant fraction of exploration improvements from novelty-based intrinsic motivation.

We prove bounds on the convergence and stability of the inverse scattering series under different Sobolev norms, finding conditions under which the radius of convergence and stability are improved.

Randomly initialized one-layer neural networks, with high probability, make datasets linearly separable.

Full waveform inversion with an optimal transport-based objective has superior convexity compared to the standard least-squares objective function for certain velocity models.