ACS 2022 Tutorial

Note

We will be adding more content to this page between now and the tutorial start. We welcome any and all questions to be asked via the Gitter public chatroom for the dcss-ai-wrapper project here: https://gitter.im/dcss-ai-wrapper/community

Updates:

  • [October 20th] First draft of this tutorial is posted online.

  • [November 20th] The tutorial will be held online using Zoom, more details to follow. See the conference slack channel # tutorial_2022

  • [November 20th] We will use the main ACS conference Zoom link

  • [November 20th] The in-person session will be in Multipurpose room 125

  • [November 21st] The tutorial has been advertised on the dcss subreddit, to attend virtually via Zoom please register here: https://docs.google.com/forms/d/e/1FAIpQLSdxlvtW95NiuozTaVz3Ie0Hbydmn-eCHnxKLGGAnPFa-alqBg/viewform

  • [November 22nd] A few quality-of-life code updates have been pushed to the master branch, you may wish to re-pull the master branch when the tutorial starts if you set up everything more than ~10 hours before the tutorial started.

Introduction

dcss-ai-wrapper aims to enable the Dungeon Crawl Stone Soup (DCSS) video game to be used as an evaluation domain for AI research. While more traditional planning benchmarks exist (i.e. IPC domains) and more traditional RL benchmarks exist (i.e. open-ai gym), it is often difficult to compare an RL agent on IPC domains or a planner on RL domains. DCSS-AI-Wrapper aims to provide support for both automated planning and RL in the DCSS video game.

Annotated Screenshot of DCSS

Annotated Screenshot of Dungeon Crawl Stone Soup’s Interface

Dungeon Crawl Stone Soup (Github) is a single-player, free, and open-source rogue-like turn-based video game that consists of a procedurally generated 2-dimensional grid world. To win the game, a player must navigate their character through a series of levels to collect `The Orb of Zot’ and then return to the starting location. Along the way, the player encounters a wide variety of monsters and items. Players equip and use items to make themselves stronger or consume them to aid in difficult situations. The DCSS environment is dynamic, stochastic, partially observable, and complex with the number of instantiated actions the player may take reaching into the hundreds.

dcss-ai-wrapper is the first AI-friendly API designed to enable planning-based agents to play Dungeon Crawl Stone Soup. In this tutorial we will introduce the DCSS game, explain the rationale and design of the API, and guide participants through exercises to run agents in DCSS. No experience with DCSS is necessary. For the hands-on portions, basic python programming experience will be useful. By the end of the tutorial you will have an automated planning agent playing DCSS on your machine using the FastDownward planner.

Schedule

The main objective of this tutorial is to provide a hands-on tutorial of the software. By the end of the tutorial, the attendees will be able to install the game / API wrapper, understand various API functionalities, and be able to run sample Automated Planning agents. We plan for a 3 hour tutorial.

Date: Tuesday, November 22nd, 2022

Time (EST): 8:45am to Noon

Location: We hope to hold the tutorial as a hybrid event, both in person at the Advances in Cognitive Systems conference and virtually through the ACS 2022 conference Zoom (you will have recieved a link to this when you registered for the conference virtually - which is free).

After the tutorial ends, you are more than welcome to continue the conversation and ask questions in the gitter chatroom for this project: https://gitter.im/dcss-ai-wrapper/community

Schedule:

Time

Description

8:45am - 9am

Setup and helping attendee’s with any installation issues

9am - 9:30am

Introduction to the DCSS Game

9:30am - 10am

DCSS-AI-Wrapper API - Rationale, current status & how you can help

10am to Noon

Excercises (organizer will be available to help guide participants through these excercises):

  • Exercise 1: Get random agent running and watch it play in the browser

  • Exercise 2: Get fastdownward agent running and watch it play in the browser

  • Exercise 3: Make a custom level

  • Exercise 4: Run the human interface agent to play the game manually

  • Exercise 5: See how far down the dungeon you can go, by modifying the planning agent

Preparing for the Tutorial

Please perform the following steps to get your system ready for the live tutorial session. This includes installing dcss-ai-wrapper and the FastDownward planner on your system.

(optional, fun) Try playing Dungeon Crawl Stone Soup yourself to get a feel for the game. See how far in the dungeon you can go!

Software prequisites:

Next install dcss-ai-wrapper by following these instructions here: Quickstart

Finally install FastDownward by following their instructions here: http://www.fast-downward.org/ObtainingAndRunningFastDownward

  • Rename the fastdownward folder to be ‘FastDownward’ and make sure it’s in the top level dcss-ai-wrapper directory, like:

    dcss-ai-wrapper/FastDownward/

  • You will need to be able to call fast-downward.py from the command line (on Windows via powershell, or in Unix based systems on the command line) so the best instructions to follow are the ones where you compile it. We have not tested using FastDownward from the Vagrant or Docker installation images.

  • If you have trouble installing FastDownward, you can still follow along with the tutorial up to a certain point, and run agents that don’t use FastDownward.

The ultimate test to see if everything is installed and working correctly is to run the fastdownwardplanningagent.py agent in src/dcss/agent/fastdownwardplanningagent.py like:

cd dcss-ai-wrapper/  # make sure you're in the project root and the virtual env is active
python .\src\dcss\agent\fastdownwardplanningagent.py

Then go to the browser and if you can see the agent moving around then it means it’s working. It should be exploring new tiles and attacking monsters whenever it finds them (albeit sometimes poorly).