ABOUT ME
I am currently a second year Ph.D. student in Electrical and Computer Engineering at the University of California San Diego, working with Professor Yang Zheng.
Previously, I was a master student at University of Michigan, Ann Arbor, majoring in Mechanical Engineering. I worked in Biped robotics lab with Professor Jessy Grizzle.
My research interests lie in control and path planning. Specifically, I’m interested in realizing safe motion planning and control for high-order systems with uncertainties (e.g., multi-agents systems, and CAVs in mixed traffic systems).
Updates
- Convex Approximations for a Bi-level Formulation of Data-Enabled Predictive Control was accepted for publication at L4DC 2024.
- Smoothing Mixed Traffic with Robust Data-driven Predictive Control for Connected and Autonomous Vehicles was accepted for publication at ACC 2024.
- Presented a talk on Convex approximations of Data-enabled Predictive Control with Applications to Mixed Traffic at CO-PI Seminars.
Recent Research
Check out my recent research works!
Willems' Fundamental Lemma for Nonlinear Systems with Koopman Linear Embedding
In this letter, we prove the data-driven representation adapted from Willems’ fundamental lemma is accurate for nonlinear systems that admit a Koopman linear embedding. The existence of the simple-to-build data-driven representation implies we can bypass the non-trivial lifting function selection process. Moreover, our derivation demonstrates the size of the trajectory library of the data-driven representation relates to the dimension of the “hidden” Koopman linear embedding of the nonlinear system.
Decentralized Robust Data-driven Predictive Control for Smoothing Mixed Traffic Flow
In this paper, we introduce a decentralized, robust DeeP-LCC (Data-Enabled Predictive Leading Cruise Control) approach to smooth traffic flow, where each CAV uses local data for control inputs. Our approach models interactions between subsystems as bounded disturbances, proposes estimation methods, and formulates a robust optimization problem with tractable solutions. Compared to centralized systems, it reduces computational load, improves safety, and maintains privacy.
Convex Approximations for a Bi-level Formulation of Data-Enabled Predictive Control
In this paper, we introduce a new bi-level formulation incorporating both system ID techniques and predictive control, and discuss how existing and new variants of DeePC can be considered as convex approximations of this bi-level formulation. Notably, a novel variant called DeeP-SVD-Iter has shown remarkable empirical performance on systems beyond deterministic LTI settings.
Past Projects
Check out my past projects here!
Comparison between experiments and two simulators with two different low-level trajectories tracking methods
In this research, we aim to discuss which simulator (MuJoCo & MATALB) is better and try to explain what causes such kind of difference. We also want to compare performances of two different tracking methods (PD & Passivity) in simulators and experiments.
Motion planning for robots with high DOFs
In this project, we aim to realize the path planning for robots with high DOFs. Also, the robot is able to avoid obstacles.
Position control of the robot endeffector
In this project, we aim to control the endeffector of the robot move to a specific position fast and accurately.
Hydraulic excavator testing model and fixed-point path planning
In this project, a testing model for hydraulic excavator is built and the path planning for automatic excavation and loading is realized by using ROS. A PID controller is used in trajectory tracking.