Dataset for "Explainable Deep Reinforcement Learning for Adaptive Patrol Speed Control of a Rail-Guided Robot System"

This repository contains the dataset used for training analysis, evaluation, and real-world validation of the proposed explainable deep reinforcement learning framework for adaptive patrol-speed control on a rail-guided robot platform.The dataset includes learning curves, real-world sensor logs, representative simulation and real-environment images, and CycleGAN domain-translation samples. These files support verification of the analyses and figures presented in the associated manuscript—including learning-curve trends, performance comparisons, Grad-CAM visualizations, and real-world sensor evaluations.However, the dataset does not include the Unity simulation environment or the full training code, and therefore does not enable end-to-end retraining of the models.