Internal Model Principle-Based Observer-LQR for Visual Servo Control of a Cartesian Ocular Robot Dataset

RESEARCH DATASET & CODEBASETitle: Internal Model Principle-Based Observer-LQR for Visual Servo Control of a Cartesian Ocular RobotAuthors: Fathih Dhiya Az Zhafran, Endra Joelianto, Faqihza Mukhlish, Arjon Turnip, Rakhmad Hidayat, Amanda Tiksnadi, Nguyen Le Hoa OVERVIEW:This repository contains the full codebase and experimental CSV datasets supporting the research paper on observer-based IMP-LQR visual servoing for a 2-DOF Cartesian ocular robot. FOLDER STRUCTURE: 1. /Code_and_Simulation/ - IPC_Main_Program_ITrack.py: Primary real-time control loop running on the IPC Harmony BX1 (Observer, LQR, PID, Deadband filter, Safety interlock). - Drowsiness_Program.py: Vision processing node for facial landmark extraction, Eye Aspect Ratio (EAR), and PERCLOS fatigue evaluation. - PSO_Program.py: Particle Swarm Optimization script for tuning LQR matrices (Q, R) and PID parameters. - Simulation_Program.py & Google_Colab-Simulation: Software-in-the-Loop (SITL) simulation scripts for benchmarking under asynchronous latency. - Camera_Calibration.py & DataGraf_Stdv_Program.py: Utilities for pinhole camera calibration and statistical plot generation (mean +- 1 SD). - /Arduino_IDE_ESP32_SensorProgram/: Microcontroller code for magnetic encoder reading. - /Arduino_IDE_Megapi_MotorProgram/: Microcontroller code for pulse translation to stepper motor drivers. - /calib_images/: Camera calibration image dataset. 2. /Experimental_Data/ - Static_Filtered_LQR_*.csv & Static_Filtered_PID_*.csv: Repeatability static station-keeping datasets with Deadband filter & Luenberger Observer enabled. - Static_Unfiltered_Run*.csv: Static station-keeping datasets without filtering (baseline noise evaluation). - Dynamic_Filtered_LQR_*.csv & Dynamic_Filtered_PID_*.csv: Dynamic trajectory tracking datasets (Step, Sine, and Unpredictable modes). - drowsiness_*.csv: Real-time operator eye aspect ratio (EAR) and PERCLOS logging data. USAGE NOTES:- Datasets are formatted in standard CSV format containing timestamps (s), sample intervals dt (s), controller modes, enc_x/y (cm), target_x/y (px), and control effort u_x/y (V).- Python scripts require Python 3.8+ with OpenCV, MediaPipe, NumPy, Matplotlib, and SciPy dependencies. CONTACT / CITATION:If you use this dataset or codebase, please cite our corresponding paper published in Taylor & Francis.