Max Robotics — Global Service Robot Directory + US Certification

About

Muhammad Zubair Irshad is a robotics and computer vision researcher whose work spans vision-and-language navigation, neural scene representations, and robotic manipulation. He is perhaps best known for his 2021 paper "Hierarchical Cross-Modal Agent for Robotics Vision-and-Language Navigation," which garnered 47 citations and introduced a hierarchical deep learning framework enabling robots to navigate complex environments using natural language instructions — a significant advancement over prior flat navigation architectures. His research has since expanded into rich 3D scene understanding, most notably through "Language-Embedded Gaussian Splats (LEGS)," a system that equips mobile robots with the ability to incrementally construct semantically aware, room-scale 3D maps in real time. His work on "NeRF-MAE" pushes the frontier of self-supervised 3D representation learning by applying masked autoencoders to neural radiance fields, demonstrating a keen interest in scalable, data-efficient learning. More recently, "ZeroGrasp" showcases his commitment to practical embodied AI, enabling zero-shot robotic grasping through shape reconstruction. Across his portfolio, Irshad consistently bridges perception, language, and action — positioning him as a rising contributor to the foundations of intelligent, autonomous robotic systems.