Code for "Robust Nonlinear System Identification using Reproducing Kernel Hilbert Spaces"

This repository contains the source code and supplementary material for the paper:

Jannis O. Lübsen and Annika Eichler, "Robust Nonlinear System Identification using Reproducing Kernel Hilbert Spaces" Submitted to the European Control Conference (ECC).

The code allows for the reproduction of all tables and figures presented in the manuscript.

---

Requirements

The code was developed and tested using MATLAB R2024b on Ubuntu 24.04.2 LTS.

Dependencies

The following third-party toolboxes are required:

• CasADi: v3.6.5
• MOSEK: v10.0.45
• Acados: v0.5.1
• GP-ML Toolbox: (Optional) Required only for the hyperparameter optimization script.

---

Installation

1. Clone Repository

2. Install Dependencies

   • CasADi & MOSEK: Install and ensure they are added to your MATLAB path.
   • Acados: Acados requires compilation. Please follow the official Acados installation guide for MATLAB.
   • GP-ML Toolbox: (Optional) Add the toolbox folder to your MATLAB path.

---

Reproducing the Results

1. To run the algorithm for the obstacle avoidance problem execute the 'run.m' script.
2. To generate the plots using the existing data go to 'plot' folder and execute 'create_plots.m'.
3. To generate new data go to 'plot->scripts_to_create_data' and run the scripts inside the folder.