Mohamed Elhoseiny

Rutgers Sexual and Reproductive Health and Rights, Meta (Israel), Kootenay Association for Science & Technology

Papers

Total Citations

H-Index

About

Mohamed Elhoseiny is a leading researcher at the intersection of computer vision, embodied AI, and human-robot interaction. His work fundamentally addresses how intelligent systems perceive, remember, and act within dynamic environments. A key contribution is his pioneering framework, "LLM as A Robotic Brain," which unifies egocentric memory and control by leveraging large language models as the central cognitive architecture for robots. This work, already garnering 10 citations since 2023, promises to simplify the traditionally separate pipelines for memory and action. Elhoseiny has also made significant strides in object recognition and pose estimation, developing a nonlinear view-invariant generative model that allows robots to recognize objects and their orientations from any viewpoint—a critical capability for manipulation. His research extends to incremental learning for video object segmentation, inspired by how children learn through human-robot interaction, and to multimodal trajectory forecasting with his "HalentNet" model, which predicts agent motion by hallucinating possible intents. With a portfolio of highly cited papers, Elhoseiny is shaping the future of embodied intelligence, making robots that are not only perceptive but also capable of continuous, interactive learning.