Max Robotics — Global Service Robot Directory + US Certification

About

Fereshteh Sadeghi is a leading robotics and machine learning researcher whose work sits at the intersection of deep reinforcement learning, sim-to-real transfer, and embodied robot control. She is best known for pioneering research on enabling robots to acquire sophisticated physical skills from simulation and transferring those capabilities to real-world systems with minimal additional training. Her landmark CAD2RL work (141 citations) demonstrated that autonomous drone navigation could be achieved using purely synthetic training data — a breakthrough that helped establish sim-to-real transfer as a cornerstone methodology in modern robotics. Building on this, her research on viewpoint-invariant visual servoing (108 citations) showed how recurrent neural networks could endow robots with human-like adaptability across diverse camera perspectives. More recently, her work on teaching bipedal robots to play soccer through deep RL (147 citations) has captured widespread attention, illustrating how complex, agile behaviors can be synthesized for humanoid platforms. Her NeRF2Real framework further advances the field by leveraging neural radiance fields to bridge the visual realism gap between simulation and deployment. Beyond cutting-edge research, Sadeghi has demonstrated a commitment to accessibility through MuSHR, an open-source robotic racecar platform designed to democratize robotics education and research. Her body of work collectively reflects a vision of robots that learn, adapt, and generalize across the messy complexity of the real world.