Ayush Shrivastava

University of Michigan–Ann Arbor, Atlanta Technical College

Papers

Total Citations

122

H-Index

About

Ayush Shrivastava is a leading researcher at the intersection of embodied AI, natural language processing, and human-robot interaction. His work focuses on enabling robots to understand and execute complex tasks through interactive dialogue, moving beyond simple instruction-following to dynamic, conversational collaboration. Shrivastava’s most significant contribution is the **TEACh** dataset (89+ citations), a landmark resource of over 3,000 human-human dialogues that allows agents to learn how to ask clarifying questions, resolve ambiguity, and recover from mistakes in real-time. This work directly addresses a critical gap in robotics: the need for machines that can converse naturally to complete tasks in human spaces. He further advanced the field with **VISITRON**, a multi-modal Transformer-based navigator that integrates visual semantics with interactive training, and tackled the practical challenge of **sim-to-real transfer** for Vision-and-Language Navigation (VLN), bridging the gap between simulated training and real-world deployment. By pioneering datasets and models that fuse language understanding with embodied action, Shrivastava is laying the groundwork for a new generation of truly interactive, helpful robots.