Gunjan Aggarwal

Papers

Total Citations

H-Index

About

Gunjan Aggarwal is a rising star in embodied AI and robotics, whose work pushes the boundaries of how virtual agents perceive and interact with open-world environments. Her primary research areas include zero-shot learning, multimodal goal embeddings, and object-goal navigation (ObjectNav). Aggarwal’s most notable contribution is the development of ZSON (Zero-Shot Object-Goal Navigation using Multimodal Goal Embeddings), a pioneering framework that enables agents to locate objects—like “find a sink”—in unfamiliar spaces without any task-specific training. This approach eliminates the need for ObjectNav-specific reward functions, making it highly scalable and applicable to real-world scenarios. With 41 citations since its 2022 publication, ZSON has quickly become a foundational reference for researchers tackling open-vocabulary navigation. Aggarwal’s work stands out for its elegant fusion of vision-language models with reinforcement learning, offering a practical path toward generalist robots. By demonstrating that agents can generalize to novel objects and environments zero-shot, she has opened new avenues for deploying AI in homes, warehouses, and beyond. For students and researchers, Aggarwal exemplifies how creative problem-solving at the intersection of language and robotics can drive impactful, real-world innovation.