Iro Armeni
Papers
5
Total Citations
35
H-Index
4
About
Iro Armeni is a leading researcher at the intersection of computer vision, robotics, and 3D scene understanding, with a particular focus on enabling intelligent systems to perceive and interact with dynamic, real-world environments. Her work tackles fundamental challenges in spatial intelligence, including robust robot control through mid-level visual representations—where her 2020 study demonstrated how structured perception can dramatically reduce the sample complexity of deep reinforcement learning for manipulation and navigation tasks. Armeni has made pioneering contributions to spatiotemporal 3D mapping, most notably through her "Nothing Stands Still" benchmark, which exposes the critical limitations of existing point cloud registration methods when faced with large geometric and temporal changes in evolving built environments. Her research on learning-based relational object matching across views advances object-level scene understanding, enabling robots to reason about tasks and interactions at a semantic level. With over 35 citations across her most influential works, Armeni's insights into construction robotics and automation have positioned her as a key voice in developing perception systems that can handle the non-static nature of human spaces—work that is foundational for long-term autonomous operation in real-world settings.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3Construction Robotics and Automation [TC Spot Light]4 citations · 2024
- 4Learning-based Relational Object Matching Across Views4 citations · 2023
- 5