Papers
20
Total Citations
382
H-Index
9
About
Irfan Essa is a prominent researcher at the intersection of robotics, computer vision, and embodied artificial intelligence, with particular expertise in reinforcement learning for navigation, semantic mapping, and human-to-robot task transfer. His most influential contribution, DD-PPO (Decentralized Distributed Proximal Policy Optimization), introduced a scalable, synchronous framework for distributed reinforcement learning that achieved near-perfect performance on PointGoal navigation tasks after training on 2.5 billion frames — garnering over 170 citations and setting a landmark benchmark in embodied AI. Complementing this work, his research on Semantic MapNet advanced the ability of robotic agents to construct allocentric spatial representations from egocentric RGB-D observations, bridging perception and scene understanding. Earlier contributions include pioneering work on linguistic transfer of assembly tasks from humans to robots and robust people-tracking from mobile platforms, demonstrating a career-long commitment to practical, deployable robotic systems. More recently, Essa has explored interactive scene exploration in cluttered environments and compositional residual learning for complex autonomous behaviors. Across his portfolio, his work consistently advances the frontier of intelligent, physically grounded agents capable of understanding and navigating the real world with increasing autonomy and sophistication.
Research Focus
Key Achievements
Top Papers
- 1DD-PPO: Learning Near-Perfect PointGoal Navigators from 2.5 Billion\n Frames171 citations · 2019
- 2
- 3Linguistic transfer of human assembly tasks to robots31 citations · 2012
- 4Decentralized Distributed PPO: Solving PointGoal Navigation15 citations · 2019
- 5
- 6
- 7Robust Tracking of People by a Mobile Robotic Agent12 citations · 1999
- 8
- 9
- 10