Osher Azulay
Papers
6
Total Citations
92
H-Index
5
About
Osher Azulay is a leading researcher at the intersection of robotic manipulation, tactile sensing, and dexterous in-hand control. His work focuses on enabling robots to perform fine, human-like manipulations by integrating high-resolution tactile feedback with learning-based control strategies. Azulay’s major contributions include the development of **AllSight**, a low-cost, high-resolution round tactile sensor that achieves zero-shot learning capability—a breakthrough that eliminates the need for per-device calibration and has already garnered 26 citations. He has also advanced **haptic-based object pose estimation** for underactuated compliant hands, addressing the challenge of occluded visual perception during manipulation. His comprehensive **survey of learning-based approaches for robotic in-hand manipulation** (18 citations) serves as a key reference for the field. Notably, Azulay has demonstrated innovative applications such as **learning to throw objects with minimal samples** using decision transformers and **SE(3)-aware object insertion** with compliant hands, both achieving 15 citations each. His recent work on augmenting tactile simulators with real-like capabilities aims to bridge the reality gap, further solidifying his impact on practical, scalable robotic dexterity.
Research Focus
Key Achievements
Top Papers
- 1
- 2Survey of learning-based approaches for robotic in-hand manipulation18 citations · 2024
- 3Haptic-Based and $SE(3)$-Aware Object Insertion Using Compliant Hands15 citations · 2022
- 4Learning to Throw With a Handful of Samples Using Decision Transformers15 citations · 2022
- 5
- 6Augmenting Tactile Simulators with Real-like and Zero-Shot Capabilities3 citations · 2024