Papers
5
Total Citations
167
H-Index
4
About
Caleb Chuck is a roboticist whose research lies at the intersection of machine learning, human-robot interaction, and manipulation. His work focuses on enabling robots to learn complex tasks from human demonstrations, particularly in challenging, real-world environments. A key contribution is his development of a hierarchy of supervisors for robot grasping in clutter (78 citations), a method that allows robots to reliably retrieve specific objects from densely packed spaces, directly relevant to applications like warehouse automation. Chuck has also critically examined how the choice of data collection strategy—comparing human-centric versus robot-centric sampling (59 citations)—affects the efficiency and robustness of deep learning from demonstrations. To further improve learning, he has pioneered statistical data cleaning techniques (17 citations) that automatically detect and correct inconsistencies in human-provided training data, reducing the burden on human teachers. More recently, in "ScrewNet" (11 citations), Chuck introduced a category-independent method using screw theory to estimate the articulation models of diverse objects like cabinets and drawers from depth images, enabling robots to interact with a wide range of everyday articulated mechanisms without prior knowledge of their specific type.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4
- 5