Liam Schramm

Rutgers, The State University of New Jersey, Rutgers Sexual and Reproductive Health and Rights

Papers

Total Citations

H-Index

About

Liam Schramm is a robotics researcher whose work sits at the intersection of machine learning, motion planning, and reinforcement learning, with a focus on enabling robots to perform complex, high-dimensional manipulation tasks. His most impactful contribution is the development of **autoregressive action sequence learning** for robotic manipulation, a universal policy architecture that treats robot actions as sequential data—akin to language modeling—allowing a single policy to generalize across diverse robots and task configurations (8 citations). Schramm also advanced sampling-based motion planning with his **Learning-Guided Exploration** framework, which uses learned priors to efficiently navigate high-dimensional spaces, addressing a long-standing bottleneck in optimal motion planning (7 citations). In reinforcement learning, he introduced **USHER**, an unbiased sampling method for Hindsight Experience Replay that improves learning from sparse rewards by correcting for sampling bias (2 citations). More recently, his work on **Diffusion-based Affordance Prediction** tackles the challenging problem of multi-modality storage, using diffusion models to predict precise 6D placements for objects into containers. Schramm’s research consistently bridges theoretical rigor with practical robotic systems, earning recognition for pushing the boundaries of what robots can learn and execute autonomously.