Papers
13
Total Citations
197
H-Index
8
About
Jens Lundell is a roboticist whose research lies at the intersection of dexterous manipulation, grasp synthesis, and scene understanding. His work is distinguished by a focus on enabling robots to handle complex, real-world scenarios—from cluttered bins to deformable objects—by integrating learning-based methods with explicit reasoning about uncertainty and physics. Lundell’s most influential contribution, “Towards Robust 6-DoF Multi-Finger Grasping in Clutter with Explicit Scene Understanding” (65 citations), set a new standard for robust grasping by combining deep learning with structured scene analysis. He has also pioneered the use of diffusion models for dexterous grasp generation (DexDiffuser, 24 citations) and advanced the field of active perception through visuo-haptic object shape completion (25 citations). His work on deformation-aware grasping and imitation learning for assembly tasks demonstrates a rare ability to bridge data-driven methods with principled robotics. With over 180 total citations and a portfolio spanning tactile sensing, POMDP planning, and navigation under uncertainty, Lundell is shaping the next generation of autonomous manipulation systems that are both intelligent and physically capable.
Research Focus
Key Achievements
Top Papers
- 1
- 2Active Visuo-Haptic Object Shape Completion25 citations · 2022
- 3DexDiffuser: Generating Dexterous Grasps With Diffusion Models24 citations · 2024
- 4Deformation-Aware Data-Driven Grasp Synthesis16 citations · 2022
- 5Imitating human search strategies for assembly15 citations · 2019
- 6Deep Network Uncertainty Maps for Indoor Navigation14 citations · 2019
- 7
- 8
- 9Constrained Generative Sampling of 6-DoF Grasps5 citations · 2023
- 10DDGC: Generative Deep Dexterous Grasping in Clutter3 citations · 2021