Projects
A readable overview of selected projects: research code, physical demos, and open repositories that make control and robotics easier to inspect, reuse, and explain.
Nonlinear optimal control (iLQR / Gauss-Newton shooting)
Iterative Gauss-Newton shooting methods for nonlinear optimal control.
-
What this project is about
A repository for implementing and testing iterative shooting methods for nonlinear optimal control, with examples built around underactuated benchmark systems.
- MATLAB implementation of iLQR / Gauss-Newton shooting workflows.
- Simulation examples for cart-pole swing-up and two-wheeled inverted-pendulum systems.
- Julia and MeshCat material for visualizing inverted-pendulum behavior.
-
Why it matters
The project is useful as a bridge between optimal-control algorithms and physical intuition: the same solver structure can be inspected in code, tested in simulation, and related to familiar unstable systems.
-
Repository
Colored Ball Balancing
A physical PID demonstrator for making feedback control visible and memorable.
-
What this project is about
A compact balancing setup built around a colored ball, where students can connect the P, I, and D terms of a controller with the response of a real physical system.
- Makes overshoot, settling time, and steady-state error easier to discuss.
- Keeps the control problem concrete, visual, and approachable for students.
- Used as an Open Days demonstrator for public-facing robotics outreach.
-
Why it matters
PID can look abstract when it is only presented as equations. This demo turns the same ideas into a visible loop: change the tuning, watch the behavior, and discuss why the system becomes faster, slower, smoother, or unstable.
-
Repository
Want a deeper dive?
If you are working on related problems, feel free to reach out. I am happy to discuss methods, implementation choices, and possible collaborations.