The Cart-Pole 4.0 P&S Course
This ETH course is a one-stop-shop tour for becoming a real control engineer. The overall goal is to swing-up and stabilize a cart-pole system. To achieve this, you will design control algorithms of increasing complexity, thereby solving increasingly challenging tasks.
While simulation environments (among others, of the cart-pole system) are popular benchmarks in machine learning, this course will provide you with hands-on experience on a real-world prototype — with all its challenges (parameter uncertainties, noise, delays, …).
Not only will you deploy and familiarize yourself with some of the most important concepts in control, but you will also be exposed to topics like state-estimation and system identification. The algorithms deployed range from PID, state-feedback including LQR, and Lyapunov-based control. If time permits, you can further dive into learning controllers, state estimation, model predictive control, or try out your own ideas.
- Workload: Six afternoons in the lab (ETL D10.1), two hours each. Additional preparation at home. Overall estimated workload: ~50 hours.
- Prerequisites: Attending the Control Systems class a-priori is recommended, otherwise the expected workload can be significantly higher. Some experience with Matlab and Simulink is recommended.
- Schedule: Biweekly, Thursday 13:00–15:00, Starting September 25, 2025
- Session 1: 25.09.2025
- Session 2: 09.10.2025
- Session 3: 23.10.2025
- Session 4: 06.11.2025
- Session 5: 20.11.2025
- Session 6: 04.12.2025
Course material: Script & Code