Papers

14

Total Citations

1,382

H-Index

11

About

Manolis Savva is a prominent researcher at the intersection of 3D scene understanding, embodied artificial intelligence, and photorealistic simulation. His work has fundamentally advanced how AI agents perceive, navigate, and interact with indoor environments. Among his most influential contributions is the Replica Dataset (2019, 384 citations), which introduced 18 highly photo-realistic 3D indoor scene reconstructions complete with semantic annotations, setting a new benchmark for indoor scene research. Equally transformative is his work on Habitat (2019, 197 citations), a high-performance simulation platform for embodied AI that has become a cornerstone tool for training virtual robots in photorealistic environments — further extended through Habitat 2.0 to support complex physics-enabled rearrangement tasks. His development of DD-PPO (171 citations) demonstrated near-perfect PointGoal navigation through massively distributed reinforcement learning across 2.5 billion frames. Savva has also contributed to physically-based rendering for scene understanding, text-driven 3D scene generation, and crowd navigation using relational graph learning. With hundreds of citations across multiple high-impact papers, his research has shaped modern embodied AI, making him an essential figure for anyone studying intelligent agent perception and simulation.

Research Focus

Key Achievements

11
H-Index
14
Papers
1,382
Total Citations
99
Avg Citations/Paper
🏆 Most Cited Paper
The Replica Dataset: A Digital Replica of Indoor Spaces
384 citations · 2019
📈 Most Prolific Year: 2019 (6 Papers)
🤝 Key Collaborators: 83
🏛 Institutions: Princeton University, Simon Fraser University, Stanford University

Top Papers

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10

Key Collaborators

Contact & Links

Available for collaboration
Content generated · 14 days ago