Tatiana Tommasi
Papers
9
Total Citations
65
H-Index
5
About
Tatiana Tommasi is a researcher whose work bridges computer vision, machine learning, and robotics, with a particular focus on 3D perception, transfer learning, and robotic manipulation. Her research addresses some of the most pressing challenges in building autonomous systems that can generalize across environments, objects, and even robot platforms. Among her most influential contributions is her work on point cloud-based robotic grasping, where her end-to-end learning framework demonstrated strong real-world performance by training directly from 3D object representations—earning 26 citations since 2022. She has also pioneered efforts in 3D open-set learning through the 3DOS benchmark, pushing the field to confront how models handle novel, unseen categories in real-world conditions. Her industrial robotics work, including PaintNet for spray painting path planning on free-form 3D surfaces, reflects a commitment to practical deployment. Earlier in her career, Tommasi explored affordance-based object recognition and cross-robot transfer learning, showing consistent interest in making perception systems more adaptable and generalizable. Her research on domain randomization for soft robot control further demonstrates her range, tackling the notoriously difficult problem of sim-to-real transfer for highly deformable systems. Collectively, her work reflects a vision of robots that learn robustly, transfer knowledge efficiently, and operate meaningfully in unstructured, real-world environments.
Research Focus
Key Achievements
Top Papers
- 1End-to-End Learning to Grasp via Sampling From Object Point Clouds26 citations · 2022
- 2
- 3
- 4Object recognition using visuo-affordance maps7 citations · 2010
- 5
- 6
- 7A Modern Take on Visual Relationship Reasoning for Grasp Planning3 citations · 2025
- 8Online vs. Offline Adaptive Domain Randomization Benchmark3 citations · 2023
- 9