Lars Berscheid
Papers
8
Total Citations
200
H-Index
4
About
Lars Berscheid is a roboticist whose work sits at the intersection of manipulation, self-supervised learning, and industrial automation. His research focuses on enabling robots to handle complex, deformable objects and perform precise tasks without explicit object models. Berscheid’s most notable contribution is **SpeedFolding** (87 citations), where he developed an efficient bimanual system for folding garments—a long-standing challenge due to the high-dimensional state space of cloth. He also pioneered self-supervised approaches for pick-and-place (75 citations), allowing robots to learn manipulation from a single demonstrated goal state, bypassing the need for pre-defined object models. In industrial settings, Berscheid advanced safe visuo-tactile feedback policies for high-tolerance insertion tasks, addressing the risks of part breakage during real-world learning. Beyond manipulation, he created **Ruckig**, an online trajectory generation algorithm respecting jerk-limited constraints, which has been adopted in real-time robotics systems. With a publication record spanning from 2018 to 2023, Berscheid’s work consistently emphasizes practical, deployable solutions—combining theoretical rigor with real-world validation. His research has garnered over 200 citations, reflecting its impact on both academic robotics and industrial applications.
Research Focus
Key Achievements
Top Papers
- 1SpeedFolding: Learning Efficient Bimanual Folding of Garments87 citations · 2022
- 2Self-Supervised Learning for Precise Pick-and-Place Without Object Model75 citations · 2020
- 3
- 4Robot Learning of Shifting Objects for Grasping in Cluttered Environments11 citations · 2019
- 5
- 6
- 7
- 8