Papers
13
Total Citations
409
H-Index
10
About
Jonathan Weisz is a roboticist whose work spans deep learning for grasping, underactuated hand design, and real-world reinforcement learning. His key research areas include robotic manipulation, tactile sensing, and reproducible benchmarking. His most influential contribution is a deep learning architecture that detects stable multi-fingered grasps directly from partial object views (123 citations), enabling robots to grasp novel objects without full 3D models. He also pioneered "blind grasping," using tactile feedback and hand kinematics to achieve stable grasps without vision (70 citations), and co-designed a highly underactuated robotic hand with three fingers controlled by a single motor (66 citations). Weisz advanced reproducible robotics through RoboBench, a platform for sharing full-system simulations (32 citations). More recently, he has applied deep reinforcement learning at scale to sort waste in office buildings with mobile manipulators (15 citations) and developed assistive grasping systems using augmented reality interfaces (14 citations). His work bridges simulation and real-world deployment, with notable achievements in underactuated hand optimization and brain-computer interfaces for grasp selection.
Research Focus
Key Achievements
Top Papers
- 1Generating multi-fingered robotic grasps via deep learning123 citations · 2015
- 2
- 3A highly-underactuated robotic hand with force and joint angle sensors66 citations · 2011
- 4RoboBench: Towards sustainable robotics system benchmarking32 citations · 2016
- 5
- 6
- 7
- 8Towards automated system and experiment reproduction in robotics14 citations · 2016
- 9Assistive grasping with an augmented reality user interface14 citations · 2017
- 10