Dimitrios Dimou
Papers
6
Total Citations
43
H-Index
3
About
Dimitrios Dimou is a robotics researcher whose work sits at the intersection of computer vision, dexterous manipulation, and machine learning. His primary research focuses on enabling robotic hands to perceive and interact with the world with human-like dexterity, tackling the fundamental challenge of grasping objects they have never seen before. His most influential work, "3DSGrasp: 3D Shape-Completion for Robotic Grasp" (23 citations), introduces a novel approach that uses 3D shape-completion to infer the full geometry of an object from sparse, incomplete point cloud data, allowing a robot to generate robust grasps even when its view is obstructed. This addresses a critical bottleneck in real-world robotic grasping. Dimou has also made significant contributions to the field of postural synergies, developing conditional generative models (e.g., variational auto-encoders) that learn a low-dimensional "synergy space" for controlling dexterous hands. His 2021 paper on this topic, with 9 citations, allows a hand to adapt its grasp based on an object's size and category. More recently, he has extended this framework to enable in-hand regrasping and force-feedback control, as seen in his 2023 and 2022 works. By combining perception with learned control, Dimou is advancing the capability of robots to perform precise, adaptive manipulation in unstructured environments.
Research Focus
Key Achievements
Top Papers
- 13DSGrasp: 3D Shape-Completion for Robotic Grasp23 citations · 2023
- 2
- 3
- 4
- 5
- 6