Max Robotics — Global Service Robot Directory + US Certification

About

Steven Morad is a robotics and artificial intelligence researcher whose work sits at the intersection of multi-robot systems, deep reinforcement learning, and embodied navigation. His research focuses on developing scalable, decentralized intelligence for robot teams — enabling autonomous agents to coordinate, navigate, and communicate without centralized control. His most influential contribution, a framework for deploying Graph Neural Network-based policies across real-world multi-robot systems (44 citations), helped bridge the gap between simulation-trained AI and physical deployment — a notoriously difficult challenge in the field. Complementing this, his NavACL method (40 citations) demonstrated that automatic curriculum learning could dramatically improve how robots learn to navigate complex real environments through vision alone. More recently, Morad has pushed toward language-conditioned multi-robot navigation, integrating Large Language Models with offline reinforcement learning to allow robot teams to follow natural language commands with minimal training data. His earlier work on planetary robotics — including path planning inside lunar and Martian lava tubes and climbing robots for low-gravity environments — reveals a researcher with an ambitious scope, driven by the vision of deploying intelligent, adaptable robots in the most challenging environments imaginable, from Earth's warehouses to the surfaces of other worlds.