Directory Description

This directory contains all code necessary to build and reproduced our experiments.

RBL-Framework

This folder contains the RBL-Framework code as found in https://github.com/martinzimmermann/RBL-Framework/releases/tag/CPS-RTSA.

gym-maze

This folder contains the original gym-maze library's Python code (https://github.com/MattChanTK/gym-maze) with the changes to enable the simulation of the grid world warehouses.

Experiments

This folder contains the code directly related to our experiments.

Building and running the experiments with docker (recommended)

Prerequisite:

• docker

Steps to build:

In the folder run the docker build command, e.g. docker build -t rbl-test:latest .. This will build everything from source into a docker image.

Steps to run the experiments:

WARNING: Executing all experiments may take a while, depending on your computer it can take more than two weeks to complete all.

After the build run the docker run command and bind the /root/Experiments/Java/Results folder of the docker container to a local folder to see the output. For example if your current folder is C:\Users\user123\Documents\RBL-Test execute docker run -v "C:\Users\user123\Documents\RBL-Test\Results:/root/Experiments/Java/Results" rbl-test:latest. The results of the experiments will then be in the Results folder on your local machine.

Building and running the experiments

Prerequisite:

• open-jdk-11
• python 3.6
• pip
• gradle 6.3

Building the experiments:

Building RBL-Framework extended with SFL reasoning (optional)

1. Change directory into the RBL-Framework folder
2. Execute gradle build
3. Replace the resulting lib /RBL-Framework/build/libs/rbl-0.5.1.jar from the experiment lib folder /Experiments/Java/lib/

Building and installing the modified gym-maze OpenAI gym environment

WARNING: This step will install the python library into your global python installation. If you don't want this, we recommend using an anaconda environment.

1. Change directory into the gym-maze folder
2. execute pip install -e .

Building the experiments

1. Open the file Experiments\Java\src\main\java\simulation\Simulation.java.
2. In line 46 change the PYTHONEXECUTABLE variable to point to your python executable. If you used anaconda make sure that the variable points to the python executable of the environment in which you installed gym-maze.
3. Change directory into Experiments\Java.
4. Execute gradle build

Running the experiments

WARNING: Executing all experiments may take a while, depending on your computer it can take more than two weeks to complete all.

1. Change directory into Experiments\Java.
2. Execute gradle run

The results are written to the folder Experiments\Java\Results