Source code for a hybrid high dimensional automated guided vehicle system design

Published: 22 March 2020| Version 1 | DOI: 10.17632/59g536yght.1
Contributors:
Mohamed RHAZZAF,
Tawfik MASROUR

Description

This code source is a simulation of a hybrid high dimensional automated guided vehicle system design model in python. In fact, our approach aims to solve the problem of dimension growth in a convex 2D environment of an Automated Guided Vehicle System (AGVS), using a Deep reinforcement learning control system of kernels with low dimensions.

Files

Steps to reproduce

the code use OpenAI gym framework that design the environment and keras rl that uses Deep reinforcement learning technics (D3QN Algorithm) So first you should install numpy, tensorflow, keras, kerasrl, and pygame by using pip command the conf.txt contain all the parameters you need to adjust, including: dimension=kernel size exploitation=exploiation (0< < 1) groupe=kernel grid size nbr_agent=number of agents in each kernel nbr_food= total number of food in the HDD-AGVS environment memory_dddqn=replay memory for D3QN mini_bach=bach size target_model_update=Target network steps folder=folder path containing statistics To run the code, you make "python Main.py conf.txt"