Max Robotics — Global Service Robot Directory + US Certification

About

Karmesh Yadav is a researcher specializing in embodied AI, robot navigation, and mobile manipulation — fields at the intersection of computer vision, robotics, and natural language understanding. His work focuses on enabling autonomous agents to perceive, reason about, and interact with real-world environments using modern AI techniques. Yadav's most impactful contribution is **OpenEQA** (2024, 46 citations), which redefines Embodied Question Answering for the foundation model era, challenging agents to develop genuine environmental understanding sufficient to answer natural language queries. His work on **image-goal navigation** (2023, 23 citations) demonstrated robust systems capable of navigating to visually specified objects in both simulation and physical environments. Through **HomeRobot** (2023, 13 citations), he contributed to open-vocabulary mobile manipulation — tackling the ambitious problem of generalized pick-and-place tasks across unseen home environments. Earlier work on preventing monocular SLAM failures using reinforcement learning (2018) reveals his foundational grounding in classical robotics. His research on last-mile visual navigation further highlights his attention to the full navigation pipeline, from exploration to precise goal localization. Collectively, Yadav's contributions are advancing the frontier of deployable, general-purpose embodied agents.