Abhishek Kadian
Papers
4
Total Citations
422
H-Index
4
About
Abhishek Kadian is a leading researcher in embodied artificial intelligence (AI), whose work has fundamentally advanced how machines perceive, navigate, and interact within 3D worlds. His primary contributions lie in developing the core infrastructure and learning algorithms that power virtual robots. Kadian is best known as a key architect of **Habitat**, a pioneering platform for embodied AI research (cited 197 times). This high-performance, photorealistic 3D simulator has become an industry standard, enabling researchers to train agents at unprecedented scale and efficiency. To solve the complex challenge of navigation within such environments, he introduced **DD-PPO** (Decentralized Distributed Proximal Policy Optimization), a scalable reinforcement learning method that achieved near-perfect PointGoal navigation after training on 2.5 billion frames (171 citations). Crucially, Kadian’s work does not stop at simulation. He is a leading voice on the **Sim2Real gap**, developing rigorous engineering tools and paradigms to measure how progress in virtual environments translates to real-world robotics (39 citations). By bridging the divide between simulation and reality, Kadian’s research provides a critical foundation for the next generation of intelligent, autonomous robots.
Research Focus
Key Achievements
Top Papers
- 1Habitat: A Platform for Embodied AI Research197 citations · 2019
- 2DD-PPO: Learning Near-Perfect PointGoal Navigators from 2.5 Billion\n Frames171 citations · 2019
- 3
- 4Decentralized Distributed PPO: Solving PointGoal Navigation15 citations · 2019