- Foundations of Robotics (Fall 2021)
Held office hours and graded exams.
- Robotics Manipulation and Locomotion (Fall 2020)
Led lab sessions and graded homework and exams.
- NYU Tandon Summer Program for Machine Learning (Summer 2020)
Co-instructed the classes, prepared course materials, and led lab sessions.
I am always looking forward to advising motivated students on independent research projects. Simply send me an Email to discuss potential project ideas.
- currently not available
Dexterous Manipulation via Model-Free Reinforcement Learning (with Jerry Wu)
The promise that robots will help us with the tedious daily tasks—folding the laundry, cleaning the dishes, or fetching a beer—relies on their ability to manipulate arbitrary objects. In this project, the student is expected to implement and study model-free reinforcement learning algorithms combined with an impedance controller, to obtain a robust policy that can not only execute a single manipulation task well, but also adapt to variations of the task through trial-and-error. The approach will be validated on pick-and-place tasks both in simulation and on real robots.
Differentiable Robot Dynamics (with Vincent Lu)
The goal of this project is to implement forward/inverse dynamics algorithms (ABA/RNEA) that are auto-differentiable with respect to the model parameters (link geometry/mass/inertia) and the generalized coordinates/velocities/accelerations using Pytorch or JAX. Such models allow convenient system identification via Gradient Descent while being able to provide uncertainty estimates when trained in ensembles. If time permits, students can also integrate such models into a reinforcement learning framework to simultaneously learn policies and interpretable robot dynamics models.