Amir Barati Farimani
Papers
11
Total Citations
55
H-Index
5
About
Amir Barati Farimani is a pioneering robotics and machine learning researcher whose work sits at the intersection of robotic manipulation, deformable object interaction, and foundation model integration. His research addresses some of the most technically demanding challenges in modern robotics, including manipulating 3D deformable materials like clay and dough, where high degrees of freedom, self-occlusion, and unpredictable deformation dynamics make traditional approaches inadequate. Through systems like SculptDiff and SculptBot, he has advanced goal-conditioned diffusion policies and pre-trained model frameworks for plastic material manipulation, while LLM-Craft and PLATO demonstrate his forward-thinking integration of large language models into robotic planning and tool use. His work on visuo-tactile pretraining (VITaL) highlights his commitment to multi-modal sensory learning for dexterous manipulation, and his OpenVR teleoperation framework addresses the persistent bottleneck of high-quality demonstration collection. Beyond manipulation, Farimani has explored energy-efficient snake robot locomotion using deep reinforcement learning. With publications accumulating citations across robotics and AI venues, his contributions are shaping next-generation autonomous systems capable of operating fluently in unstructured, real-world environments — making him an important voice in embodied intelligence research.
Research Focus
Key Achievements
Top Papers
- 1
- 2OpenVR: Teleoperation for manipulation8 citations · 2025
- 3SculptBot: Pre-Trained Models for 3D Deformable Object Manipulation8 citations · 2024
- 4
- 5
- 6
- 7
- 8
- 9
- 10PLATO: Planning with LLMs and Affordances for Tool Manipulation2 citations · 2026
Key Collaborators
Related papers
- Build on Priors: Vision--Language--Guided Neuro-Symbolic Imitation Learning for Data-Efficient Real-World Robot Manipulation
- Learning to combine primitive skills: A step towards versatile robotic manipulation §
- Learning to combine primitive skills: A step towards versatile robotic manipulation
- Imagine2Act: Leveraging Object-Action Motion Consistency from Imagined Goals for Robotic Manipulation
- Learning Robotic Manipulation through Visual Planning and Acting
Researchers in this area
Labs working in this area
If you're exploring the applied side
Commercial systems in adjacent areas you may find useful as a starting point. This is not a claim that the research here is used in them — academia and industry often move on separate tracks.
- Related commercial systems
DOBOT X-TrainerDobot · Humanoid
UR AI TrainerUniversal Robots
MagicBotMagicLab · Humanoid
Unitree Go2Unitree · Humanoid
Go2Unitree Robotics- FlashBot ArmPudu Robotics · Humanoid
- Companies in this space
Atlas RoboticsUS
Quasi RoboticsUnited States
rivelinrobotics
Suggested by topic similarity — not advertising or endorsement.