Athreyi Badithela

Papers

Total Citations

H-Index

About

Athreyi Badithela is a rising researcher at the intersection of machine learning, robotics, and computer vision, with a core focus on unsupervised learning of structured representations from visual data. Her most notable contribution is the development of **SlotGNN**, a pioneering framework that enables robots to discover multi-object representations and learn visual dynamics without human-labeled data. This work introduces two innovative architectures—SlotTransport and a graph neural network-based dynamics model—that allow agents to decompose scenes into distinct object slots and predict their future states through interaction. Although her work is still early in its trajectory, with her top-cited paper already garnering 2 citations in its first year, the conceptual leap is significant: it addresses a fundamental bottleneck in robotics, where learning object-centric models from raw pixels has traditionally required extensive supervision. By combining slot attention mechanisms with graph networks, Badithela’s approach promises to make robots more adaptable to novel environments, reducing the need for pre-programmed object knowledge. Her research is particularly impactful for students and engineers working on autonomous systems, reinforcement learning, and scene understanding, offering a scalable path toward machines that can learn the structure of the world as naturally as humans do.