Basic code and description for GoBigger challenge 2021.

Overview

GoBigger Challenge 2021

en / 中文

Challenge Description

  • 2021.11.13 We are holding a competition —— Go-Bigger: Multi-Agent Decision Intelligence Challenge. Come and make your agents in the game!

Multi-agent confrontation is an important part of decision intelligence AI, and it is also a very challenging problem. In order to enrich the multi-agent confrontation environment, OpenDILab has developed a multi-agent confrontation competitive game, named GoBigger. Based on GoBigger, the purpose of this challenge is to explore the research of multi-agent games and promote the training of technical talents in the fields of decision intelligence, so as to create a "global leading", "original" and "open" decision intelligence AI open source technology ecosystem.

This challenge needs competitors to submit their agents. We will return the score for agents to help competitors have a more correct understanding of the performance of the submitted agent. At the end of the challenge, we will fully test all submissions and the final ranking of the participating teams will be conducted.

Specific Task

This challenge uses Go-Bigger as the game environment. Go-Bigger is a multi-players competitive environment. For more details, please refer to the Go-Bigger documentation. In the match, each team participating in the challenge controls one team in the game (each team consists of multiple players). Contest participating teams need to submit an agent to control a certain team in the match and the players it contains, and obtain higher scores through teamwork, thereby achieving a higher ranking in the match.

Submission

Here in submit, we provide examples of submission for all teams in our challenge. We also provide BaseSubmission, and participants should implements their owned submission based on the code.

class BaseSubmission:

    def __init__(self, team_name, player_names):
        self.team_name = team_name
        self.player_names = player_names

    def get_actions(self, obs):
        '''
        Overview:
            You must implement this function.
        '''
        raise NotImplementedError

Note that all submission should extend with BaseSubmission. We will provide team_name and player_names for each submission as their basic parameters. team_names means the name of team which this submission controls. We also know that there are several players in a team, which is relative with the player_names in the parameters. We will call get_actions() when we try to get actions from this submission. So that participants should implements get_actions() in their submission. This function will receive obs as its parameters, which is similar with what we provide in tutorial. For example, submissions will get obs as following:

global_state, player_state = obs

global_state in details:

{
    'border': [map_width, map_height], # the map size
    'total_time': match_time, # the duration of a game
    'last_time': last_time,   # the length of time a game has been played
    'leaderboard': {
        team_name: team_size
    } # the team with its size in this game
}

Participants can find their team_name in submission matched with the team_name in leaderboard.

player_state in details:

{
    player_name: {
        'feature_layers': list(numpy.ndarray), # features of player
        'rectangle': [left_top_x, left_top_y, right_bottom_x, right_bottom_y], # the vision's position in the map
        'overlap': {
            'food': [[position.x, position.y, radius], ...], 
            'thorns': [[position.x, position.y, radius], ...],
            'spore': [[position.x, position.y, radius], ...],
            'clone': [[[position.x, position.y, radius, player_name, team_name], ...],     
        }, # all balls' info in vision
        'team_name': team_name, # the team which this player belongs to 
    }
}

However, we will only provide the submission with the player_state matched with its players. That means, if player_a and player_b (both are player name) are in the team belongs to this submission, and player_c not belongs to this team, participants will only get player_a and player_b in the submission.

After getting the obs, submissions should return actions in get_actions(). actions should look like:

{
    player_a: actions_a,
    player_b: actions_b
}

Remember that both player_a and player_b should be the name in your submission's player_names. And actions_a should be a list, which contains there items, which are the same with what we propose in action-space.

Examples and Test

We provide RandomSubmission and BotSubmission. RandomSubmission provide actions randomly, and BotSubmission provide actions based on a script. Both of them could be an example of your submission. More details in code.

We also provide an example for the pipeline of the submission. Please refer to submission_example for more details. You can also develop your agent in this directory. Once you finish your my_submission.py, you can call python -u test.py to check your submission and finally get the .tar.gz file to upload.

Supplements

If you want to add other things in your submission, such as model checkpoints or any other materials, please place them in ./supplements and tar them with submission.

Finally

You should place all your code and materials under my_submission/. Use tar zcf submission.tar.gz my_submission/ to get your final submission files. The final submission.tar.gz should be:

    - my_submission
    | - __init__.py
    | - requirements.txt
    | - my_submission.py
    | - supplements/
        | - checkpoints or other materials

Attention: __init__.py should be an empty file.

Try your first submission

Maybe you are not very familiar with our competition, but don't worry, we provide the simplest case submission! Try the following code to quickly generate a my_submission.tar.gz for submission!

$ cd submit/submission_example
$ python -u test.py

The above test.py will check whether your submission is correct. If it is correct, you will get the following output:

Success!
###################################################################
#                                                                 #
#   Now you can upload my_submission.tar.gz as your submission.   #
#                                                                 #
###################################################################

Now you only need to submit your my_submission.tar.gz!

  • Note: This submission is made of a random policy. You can check the code and change the policy to get better performance.

Submission based on DI-engine

We also develop submission_example_di based on DI-engine. You can place your ckpt in supplements to get a completed submission.

Resources

Owner
OpenDILab
Open sourced Decision Intelligence (DI), powered by SenseTime X-Lab & Shanghai AI Lab
OpenDILab
Daily knowledge pills to get better in Python.

Python daily pills Daily knowledge pills to get better Python code. Why Does your Python code suffers of any of this symptoms? Incorrect Indentation I

Jeferson Vaz dos Santos 35 Sep 19, 2022
A collection of simple tools that proved to be needed for hadling large periodic calculations with the VASP software package.

VESTA-tools A collection of simple tools that proved to be needed for handling large periodic calculations with the VASP software package. distTotCalc

Ilia Kichev 2 Dec 14, 2021
The mock Pokemon Environment I built in 2019 to study Reinforcement Learning + Pokemon

ghetto-pokemon-rl-environment ##NOT MAINTAINED! Fork and maintain yourself. Environment I made back in 2019 to use Pokemon to practice reinforcement l

2 Dec 09, 2021
InfiniPy has some neat features - like the endpoint for function

InfiniPy has some neat features - like the endpoint for function

ZeroTwo 7 Nov 20, 2022
Craxk is a SINGLE AND NON-REPLICABLE Hash that uses data from the hardware where it is executed to form a hash that can only be reproduced by a single machine.

What is Craxk ? Craxk is a UNIQUE AND NON-REPLICABLE Hash that uses data from the hardware where it is executed to form a hash that can only be reprod

5 Jun 19, 2021
Beancount Importers for DKB (Deutsche Kredit Bank) CSV Exports

Beancount DKB Importer beancount-dkb provides an Importer for converting CSV exports of DKB (Deutsche Kreditbank) account summaries to the Beancount f

Siddhant Goel 24 Aug 06, 2022
Access Modbus RTU via API call to Sungrow WiNet-S

SungrowModbusWebClient Access Modbus RTU via API call to Sungrow WiNet-S Class based on pymodbus.ModbusTcpClient, completely interchangeable, just rep

8 Oct 30, 2022
Refer'd Resume Scanner

Refer'd Resume Scanner I wanted to share a free resource we built to assist applicants with resume building. Our resume scanner identifies potential s

Refer'd 74 Mar 07, 2022
Now you'll never be late for your Webinars or Meetings on the GoToWebinar Platform

GoToWebinar Launcher : Now you'll never be late for your Webinars or Meetings on the GoToWebinar Platform About Are you popular for always being late

Jay Thorat 6 Jun 07, 2022
Python-Roadmap - Дорожная карта по изучению Python

Python Roadmap Я решил сделать что-то вроде дорожной карты (Roadmap) для изучения языка Python. Возможно, если найдутся желающие дополнять ее, модифиц

Ruslan Prokhorov 1.2k Dec 28, 2022
This Curve Editor, written by Jehee Lee in 2015

Splines Abstract This Curve Editor, written by Jehee Lee in 2015, is a freeware. You can use, modify, redistribute the code without restriction. This

Movement Research Lab 8 Mar 11, 2022
A project to work with databases in 4 worksheets, insert, update, select, delete using Python and MySqI

A project to work with databases in 4 worksheets, insert, update, select, delete using Python and MySqI As a small project for school or college hope it is useful

Sina Org 1 Jan 11, 2022
Python Excuse Generator

Excuse Generator Python Excuse Generator This project is an excuse generator that provides the user with an excuse as to why they weren't paying atten

Collin Sanders 5 Jul 07, 2022
A PDM plugin to publish to PyPI

PDM Publish A PDM plugin to publish to PyPI NOTE: Consider if you need this over using twine directly Installation If you installed pdm via pipx: pipx

Branch Vincent 20 Aug 06, 2022
Simple package to make requests throughout Tor with circuit renewal.

AutoTor Table of Contents About the Project Contents Dependencies Getting Started Installation Coding Contributing About the Project Simple package to

Salvador Belenguer 6 Jan 01, 2023
Write complicated anonymous functions other than lambdas in Python.

lambdex allows you to write multi-line anonymous function expression (called a lambdex) in an idiomatic manner.

Xie Jingyi 71 May 19, 2022
my own python useful functions

PythonToolKit Motivation This Repo should help save time for data scientists' daily work regarding the Time Series regression task by providing functi

Kai 2 Oct 01, 2022
Lightweight Scheduled Blocks Checker for Current Epoch. No cardano-node Required, data is taken from blockfrost.io

ReLeaderLogs For Cardano Stakepool Operators: Lightweight Scheduled Blocks Checker for Current Epoch. No cardano-node Required, data is taken from blo

SNAKE (Cardano Stakepool) 2 Oct 19, 2021
An animal facts python module

An animal facts python module

Fayas Noushad 3 Dec 19, 2021
Perform oocyst segmentation in mercurochrome stained mosquito midgut

Midgut_oocyst_segmentation Perform oocyst segmentation in mercurochrome stained mosquito midguts This oocyst segmentation model also powers the webtoo

Duo Peng 3 Oct 27, 2021