A python interface for training Reinforcement Learning agents to play the Chef's Hat Card Game.
Project description
ChefsHatGym V2
This repository holds the ChefsHatGym2 environment, which contains all the necessary tools to run, train and evaluate your agents while they play the Chef`s Hat game.
With this library, you will be able to:
- Encapsulate existing agents into the game
- Run the game locally, on your machine
- Connect the game to remote agents (using a pub/sub architecture powered by a Redis server)
- Export experimental results, game summaries and agents behavior on a easy-to-read format
- Evaluate agents using different evaluation tools and visualizations
Full documentation can be found here: Documentation.
We also provide a list of existing plugins and extensions for this library:
Chef`s Hat Players Club
The Chef’s Hat Player’s Club is a collection of ready-to-use artificial agents. Each of these agents were implemented, evaluated, and discussed in specific peer-reviewed publications and can be used at any time. If you want your agent to be included in the Player’s Club, send us a message.
Chef`s Hat Play With a Human Plugin
Comming soon...
The Chef's Hat Card game
The Chef's Hat Environment provides a simple and easy-to-use API, based on the OpenAI GYM interface, for implementing, embedding, deploying, and evaluating reinforcement learning agents.
Fora a complete overview on the development of the game, refer to:
- It's Food Fight! Introducing the Chef's Hat Card Game for Affective-Aware HRI (https://arxiv.org/abs/2002.11458)
- You Were Always on My Mind: Introducing Chef’s Hat and COPPER for Personalized Reinforcement Learning (https://www.frontiersin.org/articles/10.3389/frobt.2021.669990/full)
- The Chef's Hat rulebook The Chef's Hat rulebook.
If you want to have access to the game materials (cards and playing field), please contact us using the contact information at the end of the page.
Summary of game rules
During each game there are three phases: Start of the game, Making Pizzas, End of the game. The game starts with the cards been shuffled and dealt with the players. Then, starting from the second game, the exchange of roles takes place based on the last games' finishing positions. The player who finished first becomes the Chef, the one that finished second becomes the Sous-Chef, the one that finished third becomes the Waiter and the last one the Dishwasher. Once the roles are exchanged, the exchange of the cards starts. The Dishwasher has to give the two cards with the highest values to the Chef, who in return gives back two cards of their liking. The Waiter has to give their lowest valued card to the Sous-Chef, who in return gives one card of their liking.
If, after the exchange of roles, any of the players have two jokers at hand, they can perform a special action: in case of the Dishwasher, this is "Food Fight" (the hierarchy is inverted), in case of the other roles it is "Dinner is served" (there will be no card exchange during that game).
Once all of the cards and roles are exchanged, the game starts. The goal of each player is to discard all the cards at hand. They can do this by making a pizza by laying down the cards into the playing field, represented by a pizza dough. The person who possesses a Golden 11 card at hand starts making the first pizza of the game. A pizza is done when no one can, or wants, to lay down any ingredients anymore. A player can play cards by discarding their ingredient cards on the pizza base. To play cards, they need to be rarer (i.e. lowest face values) than the previously played cards. The ingredients are played from highest to the lowest number, that means from 11 to 1. Players can play multiple copies of an ingredient at once, but always have to play an equal or greater amount of copies than the previous player did. If a player cannot (or does not want) to play, they pass until the next pizza starts. A joker card is also available and when played together with other cards, it assumes their value. When played alone, the joker has the highest face value (12). Once everyone has passed, they start a new pizza by cleaning the playing field, and the last player to play an ingredient is the first one to start the new pizza.
Chef`sHatGym2 Simulator
Instalation
You can use our pip installation:
pip install chefshatgym
To use the remote communication, you need access to a redis server. To install your own redis server, please follow: https://redis.io/docs/getting-started/installation/
For Windows users, Memurai is an alternative.
Refer to our full documentation for a complete usage and development guide.
Running a game locally
The basic structure of the simulator is a room, that will host four players, and initialize the game. ChefsHatGym2 encapsulates the entire room structure, so it is easy to create a game using just a few lines of code:
# Start the room
room = ChefsHatRoomLocal(
room_name="local_room",
verbose=False,
)
# Create the players
p1 = AgentRandonLocal(name="01")
p2 = AgentRandonLocal(name="02")
p3 = AgentRandonLocal(name="03")
p4 = AgentRandonLocal(name="04")
# Adding players to the room
for p in [p1, p2, p3, p4]:
room.add_player(p)
# Start the game
info = room.start_new_game(game_verbose=True)
For a more detailed example, check the examples folder.
Running a game remotely
ChefsHatGym2 allows for communication with remote agents. It uses a pub/sub architecture, powered by a Redis server. A remote room structure is provided by the library, as shown in our examples folder.
Chefs Hat Agents
ChefsHatGym2 provides an interface to encapsulate agents. It allows the extension of existing agents, but also the creation of new agents. Implementing from this interface, allow your agents to be inserted in any Chef`s Hat game run by the simulator.
Runing an agent from another machine is supported, by the ChefsHatRemote agent interface.
Here are examples of an agent that only select random actions, implementing both local and remote interfaces:
Legacy Plugins and Extensions
Chef's Hat Online (ChefsHatGymV1)
The Chef’s Hat Online encapsulates the Chef’s Hat Environment and allows a human to play against three agents. The system is built using a web platform, which allows you to deploy it on a web server and run it from any device. The data collected by the Chef’s Hat Online is presented in the same format as the Chef’s Hat Gym, and can be used to train or update agents, but also to leverage human performance.
Moody Framework (ChefsHatGymV1)
Moody Framework is a plugin that endowes each agent with an intrinsic state which is impacted by the agent's own actions.
Use and distribution policy
All the examples in this repository are distributed under a Non-Comercial license. If you use this environment, you have to agree with the following itens:
- To cite our associated references in any of your publication that make any use of these examples.
- To use the environment for research purpose only.
- To not provide the environment to any second parties.
Citations
-
Barros, P., Yalçın, Ö. N., Tanevska, A., & Sciutti, A. (2023). Incorporating rivalry in reinforcement learning for a competitive game. Neural Computing and Applications, 35(23), 16739-16752.
-
Barros, P., & Sciutti, A. (2022). All by Myself: Learning individualized competitive behavior with a contrastive reinforcement learning optimization. Neural Networks, 150, 364-376.
-
Barros, P., Yalçın, Ö. N., Tanevska, A., & Sciutti, A. (2022). Incorporating Rivalry in reinforcement learning for a competitive game. Neural Computing and Applications, 1-14.
-
Barros, P., Tanevska, A., & Sciutti, A. (2021, January). Learning from learners: Adapting reinforcement learning agents to be competitive in a card game. In 2020 25th International Conference on Pattern Recognition (ICPR) (pp. 2716-2723). IEEE.
-
Barros, P., Sciutti, A., Bloem, A. C., Hootsmans, I. M., Opheij, L. M., Toebosch, R. H., & Barakova, E. (2021, March). It's Food Fight! Designing the Chef's Hat Card Game for Affective-Aware HRI. In Companion of the 2021 ACM/IEEE International Conference on Human-Robot Interaction (pp. 524-528).
-
Barros, P., Tanevska, A., Cruz, F., & Sciutti, A. (2020, October). Moody Learners-Explaining Competitive Behaviour of Reinforcement Learning Agents. In 2020 Joint IEEE 10th International Conference on Development and Learning and Epigenetic Robotics (ICDL-EpiRob) (pp. 1-8). IEEE.
-
Barros, P., Sciutti, A., Bloem, A. C., Hootsmans, I. M., Opheij, L. M., Toebosch, R. H., & Barakova, E. (2021, March). It's food fight! Designing the chef's hat card game for affective-aware HRI. In Companion of the 2021 ACM/IEEE International Conference on Human-Robot Interaction (pp. 524-528).
Events
Chef`s Hat Cup: Revenge of the Agent!
Get more information here: https://www.chefshatcup.poli.br/home
The First Chef's Hat Cup is online!
Get more information here: https://www.whisperproject.eu/chefshat#competition
Contact
Pablo Barros - pablovin@gmail.com
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for ChefsHatGym-2.0.2-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | fc3ff6fd4547337a91a19e3f9647ad2ef64ef2f6f00551d0785d773c47066c8b |
|
MD5 | a4eb19a1ce4fdcdbfc5aec0a3d001eb7 |
|
BLAKE2b-256 | 81ff554426db90ec00900d6cac9c67f3cfb5ed949c6cf5a06d7a62dafcc71470 |