iPendulum

Overview

This repository is a MATLAB / Simulink learning lab for inverted-pendulum control. It explores classical and modern control methods on the cart-pole benchmark, including PID, cascade PID, pole placement, LQR, LQG, Kalman filtering, and Luenberger observer examples.

The repo is best understood as a concrete companion to robotics-control-learning-labs: the pinned control hub explains the broader control-learning direction, while this repo provides a focused inverted-pendulum application.

Repository Role

This repo supports the public robotics/control portfolio by showing applied control practice on a standard nonlinear benchmark. It connects to the broader research direction through state-space modeling, observer design, controller comparison, tracking/regulation metrics, and MATLAB/Simulink simulation workflows.

It should be treated as a public support artifact, not the main PhD research repository. Unpublished uncertainty-aware control work, paper-specific ablations, and private experiment notes should stay private elsewhere.

Research/Engineering Motivation

The inverted pendulum is a classic benchmark for learning feedback control because it combines instability, nonlinear dynamics, actuator effort, state estimation, and controller tuning in a compact system.

For mobile manipulation research, these control concepts remain useful because scan execution and robot motion depend on stable tracking, observers, disturbance handling, and meaningful comparisons between baseline controllers.

Features

• MATLAB and Simulink inverted-pendulum experiments.
• PID and cascade PID baseline simulations.
• LQR state-feedback examples.
• Pole-placement and Luenberger observer examples.
• LQG / Kalman-filter style simulation workflow.
• Animated demo GIFs for selected controllers.
• Modular utilities under src/ and scenario runners under sim/.

Method

The repository follows a control-learning workflow:

1. Define or load inverted-pendulum system parameters.
2. Build a state-space model for the cart-pole dynamics.
3. Design a controller or observer.
4. Run a simulation scenario.
5. Compute basic control metrics.
6. Plot or animate the response.
7. Compare controller behavior and limitations.

Installation

Clone the repository and add it to the MATLAB path:

git clone https://github.com/WikiGenius/ipendulum.git
cd ipendulum

In MATLAB:

addpath(genpath(pwd))

Run

Example simulations:

sim/run_pid_simulation
sim/run_pid_cascade_simulation
sim/run_lqr_kalman_simulation
sim/run_place_luenberger_simulation

The exact requirements depend on the simulation or Live Script being used.

Results

Example demo GIFs are stored under docs/:

 

Additional media includes:

docs/ipendulum_demo.gif
docs/demo_cheap_states_LQG.mp4
docs/demo_polePlacement.mp4

Project Structure

.github/workflows/   MATLAB-oriented workflow example
data/                system parameters and setup scripts
docs/                demo GIFs, videos, and documentation media
notebooks/           MATLAB Live Scripts for usage examples
sim/                 scenario runners for PID, LQR/LQG, and observer demos
src/                 reusable modeling, plotting, and metric utilities

Requirements

• MATLAB R2021b or newer is recommended.
• Simulink may be required for some workflows.
• Control System Toolbox may be required for controller/observer design functions.

Limitations

• This is a learning/control lab, not a production controller.
• Some workflows depend on MATLAB Live Scripts and toolbox availability.
• System parameters and demos are intended for reproducible study, not hardware deployment.
• This repo should not contain unpublished PhD algorithms, advisor notes, or private ablation studies.
• The current README documents the public-safe learning value; future work can add clearer metrics tables and more polished result plots.

Roadmap

• Add a short state-space derivation note.
• Add a controller comparison table.
• Add reproducible metric outputs under results/.
• Add a public-safe LQR tuning note linked to robotics-control-learning-labs.
• Clean or remove generated metadata files that are not useful to readers.
• Add a clear rights/license file if this repo is released publicly as open-source.

Citation / Acknowledgment

This repository uses standard inverted-pendulum control concepts from classical control, state-space control, optimal control, and estimation. Cite any textbook, course, MATLAB example, or library used when extending the examples.

Rights and Reuse

This repository is shared for academic portfolio and demonstration purposes. Unless a separate open-source license is added, all rights are reserved by the author.