Robot Learning Research
2026 — Present
Imitation learning pipelines that let robotic manipulators mimic human dexterity from video input.
PythonFoundationPoseSAM36D Pose EstimationPyTorch
At the Loyola Chicago Software Systems Laboratory, I'm investigating imitation learning pipelines that enable robotic manipulators to mimic human dexterity directly from video demonstrations.
The work combines state-of-the-art foundation models for perception with custom motion retargeting so that a human's hand trajectory can be mapped onto the constraints of real robot hardware.
Highlights
- Integrating foundation models (FoundationPose, SAM3) for robust 6D pose estimation and object segmentation.
- Developing motion retargeting algorithms that map human hand trajectories onto robot hardware constraints.