Changhyun Choi
University of Minnesota, MN·Computer Science·Early-career
People can easily grab a specific item from a crowded drawer, slide a flat object against a box wall to pick it up, or dig through a pile to find something hidden. Robots still struggle to grasp objects in anything beyond moderately cluttered tabletops.
- Fields
Computer Science
Engineering
- Topics
robotics
robotic grasping
robotic manipulation
- Grants
- 1 known grant, through May 2027
What they work on
- Develop context-aware grasping algorithms that let robots consider clutter, reachability and collisions when picking objects from crowded scenes, using visual scene understanding and self-supervised learning.
- Enable robots to grasp hard-to-pick objects by using fixtures such as walls and corners, learning pixel-level or object-level affordances (which parts of a scene support an action) from camera images.
- Create methods for robots to search for and grasp a hidden target object in a pile, given a reference image or a natural-language request, while learning to adapt to new objects and environments with minimal human guidance.
More about the research
Grasping is a first step for nearly every manipulation task, so this limit keeps robots out of many settings where they could help.
Why it matters
Robots that can reliably grasp objects in cluttered or confined spaces could help in places where robot labor is not yet practical. Examples include handing parts to factory workers, harvesting fruits and vegetables, picking items from warehouse or grocery shelves, and fetching items like a remote control or pills in eldercare. This work aims to lay groundwork for affordable, dependable robotic help in homes, factories, farms and care facilities.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Shuran Song
Stanford University, CA·Computer Science·Early-career
Develop hierarchical task representations, built in hyperbolic space (a tree-like curved geometry), learned from large-scale unlabeled instructional videos to model the steps of everyday activities such as cooking and repair.
Xiaolong Wang
University of California-San Diego, CA·Computer Science·Early-career
Recognize objects in videos of everyday environments without a fixed category list, training with text supervision and self-supervision instead of detailed labels.
Noah Snavely
Cornell University, NY·Computer Science
Develop compositional implicit neural scene representations that pair a network's ability to capture fine detail with physical models of scene structure, such as light fields.
The school
University of Minnesota: 88 labs on PI Pages, $1.4B a year on research, 722 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.