Prithviraj Ammanabrolu
Prithviraj Ammanabrolu is a tenure-track assistant professor in the Department of Computer Science and Engineering at UC San Diego, where he leads the PEARLS Lab, and concurrently serves as a research scientist at NVIDIA. He received his PhD from Georgia Institute of Technology, where he was advised by Mark Riedl, and previously worked at the Allen Institute for AI and MosaicML. He is best known for research at the intersection of language grounding and embodied AI, focusing on language-enabled reinforcement learning and the development of AI agents that interactively align to human preferences in grounded environments. His recent work includes DeMiAn, a framework for dense multi-aspect language annotation of robot demonstrations to improve policy learning without collecting new data.
“DeMiAn: Dense Multi-Aspect Annotation for Robot Policy Learning”
Source→“On RoboCasa, the best fixed annotation type (Physical Motion) raised success rate from 44% to 46%, and the learned instructor pushed that to 49% — within 3 points of a theoretical per-task oracle at 52%.”
Source→“Async tracks sync within fractional points on SR (49.0% vs 49.5%) while injecting the instruction into a rollout already in progress.”
Source→“Senior author and a recognized researcher at the intersection of language grounding and embodied AI. His work on language-conditioned agents is directly foundational to DeMiAn's framing.”
Source→AI-extracted from podcast / newsletter / paper summaries. May contain errors.