Mark Van der Merwe
Papers
5
Total Citations
76
H-Index
4
About
Mark Van der Merwe is a leading researcher in robotic manipulation, with a focus on integrating perception, learning, and control for dexterous grasping and tool use. His major contributions include pioneering deep learning methods for multifingered grasp planning, where he demonstrated that learned neural network models can outperform traditional sampling-based approaches by predicting grasp success from 3D visual data (56 citations). He has also advanced geometrically aware grasping by developing continuous 3D reconstruction techniques that enable robots to reason about full object geometry from partial views. Van der Merwe’s work on visuo-tactile transformers introduced novel multimodal representation learning that fuses vision and touch to improve manipulation dexterity and robustness. His research on compliant tool-environment interaction, such as scraping with a spatula, has pushed the boundaries of contact servoing. Most recently, his This&That framework (2025) integrates language and gesture control for video generation to guide robot planning. With a growing citation impact and a focus on practical, real-world manipulation, Van der Merwe is shaping the future of intelligent robotic systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2Learning Continuous 3D Reconstructions for Geometrically Aware Grasping7 citations · 2020
- 3
- 4
- 5Visuo-Tactile Transformers for Manipulation4 citations · 2022