Hardware-Aware Nussbaum-PID Controller on the Niryo NED3 Pro

Companion code repository for the manuscript:

Hardware-Aware Experimental Assessment of a Nussbaum-Function PID Controller on a Low-Cost Manipulator Joint Using Optuna-Guided Tuning Danial Zafaranchizadeh Moghaddam, Abolfazl Zaraki — University of Hertfordshire, 2026.

This repo collects the public reproducibility material that backs the claims of the paper:

• the publication figures used in the article (figures/),
• the publication tables (tables/),
• the file names and documentation for the Python scripts used to regenerate the figures (scripts/).

The manuscript text, LaTeX source, and the published PDF are not distributed here. The article, its DOI, and a direct PDF link are available from the project landing page:

Project page: https://danielz.co.uk/projects/hardware-aware-nussbaum-pid/

If you use the public figures, tables, or experimental methodology, please cite the manuscript above and link back to the project page. For full code access, please email the authors listed below.

.
├── figures/    Publication figures (PDF / PNG / SVG, exact versions used in the paper)
├── tables/     Publication tables (selected metrics, parameter set, Optuna campaigns)
├── scripts/    Reproducible figure-generation Python scripts
│   ├── README.md
│   ├── build_main_results_figures.py
│   ├── build_envelope_figures.py
│   └── build_control_architecture_diagram.py
├── data/       Raw closed-loop CSV logs, per-run metrics (CSV+JSON),
│   │          Optuna trial archives, controller source code, run commands
│   ├── README.md
│   ├── best_results/{raw,metrics,plots}/
│   ├── optuna_trials/
│   ├── envelope_step_bandwidth/
│   ├── controller_code/
│   └── run_commands/
├── LICENSE
└── README.md   This file

All scripts are run from the project root and write into figures/ and tables/.

Dependencies:

python3 -m venv .venv && source .venv/bin/activate
pip install numpy pandas matplotlib pymupdf pillow
python3 scripts/build_envelope_figures.py
python3 scripts/build_main_results_figures.py
python3 scripts/build_control_architecture_diagram.py

The raw closed-loop CSV logs, per-run metric tables (CSV + JSON), Optuna trial archives, file names for controller code, and run commands are listed under data/ (see data/README.md). The PDF / PNG / SVG figures already committed to figures/ are the exact versions used in the published article and require no re-build. Python implementation bodies are not included in the public copy.

All real-hardware runs were executed on a Niryo NED3 Pro manipulator, driving Dynamixel ID 6 (Niryo J5) in raw current mode (Mode 0).

Validated envelope:

• Sinusoid: 10° and 40° amplitudes, 0.05 Hz and 0.5 Hz frequencies, locked-centre around 140°, 300 s validation.
• Step: 15° and 40° down-steps from the 140° operating centre.
• Bandwidth probes (out-of-envelope, characterized but not claimed as in-envelope tracking): 10°/1.5 Hz, 10°/3 Hz, 40°/1.0 Hz, 40°/1.5 Hz.

The public repository materials are released under the MIT License where applicable (see LICENSE). The publication figures themselves are © the authors; please cite the paper if you reuse them in your own work. Full implementation code is available by request.

• Danial Zafaranchizadeh Moghaddam — danial.za@outlook.com
• Abolfazl Zaraki — a.zaraki@herts.ac.uk

This public repository keeps the project page, file names, figures, tables, and README descriptions visible for manuscript context. Python implementation bodies are not included in the public copy.

For full code access, reviewer material, or collaboration details, please email:

• Danial Zafaranchizadeh Moghaddam — danial.za@outlook.com
• Abolfazl Zaraki — a.zaraki@herts.ac.uk

Please include your affiliation, intended use, and whether the request is for review, reproduction, or collaboration.