Implementation and simulation in MATLAB/Simulink of an Iterative Learning Control (ILC) algorithm for trajectory tracking of a 2-DOF planar manipulator. The repository also includes kinematics and dynamics for a 6-DOF anthropomorphic arm with spherical wrist, developed during the thesis work.

Context: Bachelor thesis in Computer Engineering — Algoritmo di apprendimento iterativo per braccio robotico (Roma Tre University, A.Y. 2018/2019).

---

Quick navigation

• Overview
• Project structure
• Results & media
• Getting started
• Main components
• Requirements
• Thesis & references

---

Overview

Module	Description
Planar 2-DOF	Dynamics (Lagrange), ILC and PD control, circular trajectory generation and simulation.
Kinematics	Denavit–Hartenberg utilities, direct/inverse kinematics for 2-DOF, 3-DOF and 6-DOF serial arms.
6-DOF arm	Anthropomorphic arm with spherical wrist: DH parameters, forward/inverse kinematics.

---

Project structure

ILC_2DOF_PlanarArm/
├── assets/               # Output images and videos (ILC/PD results, animations)
├── matlab/
│   ├── kinematics/       # DH matrix, dk03, dk06, ik02, ik03, ik06, ik36
│   ├── planar_2dof/      # manipulator2dof, ILC2dof, PD2dof, Traiettoria_Circolare_Corretta
│   └── six_dof/          # SIXDOF_DHparameters
├── simulink/             # Simulink models (controller2.slx, robot2link)
├── docs/                 # Thesis PDF and extra documentation
├── startup.m             # Adds matlab/ to path (run from project root)
└── README.md

---

Results & media

Outputs from the ILC and PD simulations (figures and animations) are stored in assets/.

ILC — output position and error

Link 1 position	Link 2 position	Position error (Link 1)	Position error (Link 2)

PD control — comparison

Link 1 position (PD)	Link 2 position (PD)	Position error L1 (PD)	Position error L2 (PD)

Control signals and workspace

Control position L1	Control position L2	6-DOF workspace

Animations

ILC — manipulator motion vs desired trajectory over iterations:

PD — manipulator motion with proportional-derivative control:

---

Getting started

1. Clone the repository and open MATLAB.
2. From the project root, run:

   startup

   This adds all matlab/ subfolders and simulink/ to your path (so controller2.slx is found).
3. Run scripts from the project root :
   • Circular trajectory only: matlab/planar_2dof/Traiettoria_Circolare_Corretta
   • 2-DOF dynamics (Lagrange, state-space): matlab/planar_2dof/manipulator2dof
   • ILC control (uses simulink/controller2.slx): matlab/planar_2dof/ILC2dof
   • PD control (uses controller2.slx): matlab/planar_2dof/PD2dof
   • 6-DOF arm (DH + visualization): matlab/six_dof/SIXDOF_DHparameters

Note: ILC2dof and PD2dof call sim('controller2'). The model controller2.slx is in simulink/; startup adds that folder to the path when you run from the project root.

---

Main components

Kinematics (matlab/kinematics/)

File	Description
DHmatrix.m	Builds 4×4 DH transformation from a single row of parameters.
ik02.m	Inverse kinematics for 2-DOF planar arm (position → joint angles).
dk03.m	Direct kinematics for 3-DOF arm (joint angles → end-effector position).
dk06.m	Direct kinematics for 6-DOF arm.
ik03.m, ik06.m, ik36.m	Inverse kinematics for 3-DOF and 6-DOF (arm + wrist).

Planar 2-DOF (matlab/planar_2dof/)

File	Description
manipulator2dof.m	Lagrange equations, state-space form, linearization.
Traiettoria_Circolare_Corretta.m	Circular trajectory in Cartesian space and joint space (uses ik02).
ILC2dof.m	ILC algorithm with circular reference; iteratively updates control input.
PD2dof.m	PD control on the same 2-DOF setup (single run).

6-DOF (matlab/six_dof/)

File	Description
SIXDOF_DHparameters.m	Builds 6-DOF robot with DH parameters, solves inverse kinematics and visualizes.

---

Requirements

• MATLAB (tested with Robotics System Toolbox for RigidBodyTree and related functions).
• Simulink for running ILC2dof and PD2dof (they call the controller2 model).

---

Thesis & references

• Thesis PDF: docs/tesi.pdf — Algoritmo di apprendimento iterativo per braccio robotico, Roma Tre, A.Y. 2018/2019.
• ILC updates the control signal from trial to trial using position and velocity error (e.g. ( u_{k+1} = u_k + \phi, e_{\text{pos}} + \gamma, e_{\text{vel}} )).

---

_{Bachelor thesis project — Roma Tre University · Computer Engineering}