COG 2023 Football AI Competition

Background

The Conference on Games Football AI Competition use the multi-agent scenarios of the Google Research Football (GRF) simulators as the testbed. The GRF simulator bears great similarity to well-known football games like FIFA and REAL Football, vividly modelling the dynamics of the ball and player movement as well as their interaction such as passing, shooting, etc. The competition focuses on the full-game scenario built on GRF simulator where two teams of players compete for goals on the pitch.

The competition will be held on Jidi platform which offers online evaluation of submitted policies on various simulated environments and holds Kaggle-like customized competitions. Regarding the scale of the game, the GRF competition has two tracks as two separate contests, one easy track using the 5vs5 multi-agent full-game scenario and one hard track using the 11vs11 multi-agent full-game scenario.

Tracks

Track - 1: 5_vs_5 scenario

In the 5_vs_5 multi-agent scenario, each team has four controllable player plus one goal-keeper controlled by the built-in AI. The duration of the game is 3000 discrete time-steps cooresponding to 90 minutes in a real football match. The observation for each player contains information of both the teammates and the opponents, as well as the state of the ball. Based on the provided information, each individual player can either move or make specific actions such as ball-passing or ball-shooting. The goal of each team is to control all of its players, organize team offense and defense and win the game by scoring more than the other team. Expectedly, a strong participant is required to demonstrate sophisticated football skills and counter opponents' tactics in such a complex competitive tasks.

Track - 2: 11_vs_11 scenario

In the 11_vs_11 multi-agent scenario, each team has ten controllable player plus one goal-keeper controlled by the built-in AI. The duration of the game is 3000 discrete time-steps cooresponding to 90 minutes in a real football match. The observation for each player contains information of both the teammates and the opponents, as well as the state of the ball. Based on the provided information, each individual player can either move or make specific actions such as ball-passing or ball-shooting. The goal of each team is to control all of its players, organize team offense and defense and win the game by scoring more than the other team. Expectedly, a strong participant is required to demonstrate sophisticated football skills and counter opponents' tactics in such a complex competitive tasks.

Prize

Champion of Track 2

$ 500

(sponsored by the IEEE CIS Education Competition Subcommittee)

Participation

Come and Sign up for the competition!

We provide Jidi tutorial and source code for the competition

Timeline

Open Registration: 2023.04.01, 10:30 (UTC+8)
Warm-up 1st Round (Submission Deadline 2023.06.28 10:30 UTC+8): Swiss Round double competition system, the result is not included in the final competition result
Warm-up 2st Round (Submission Deadline 2023.07.12 10:30 UTC+8): Swiss Round double competition system, the result is not included in the final competition result
Final 1st Round (Submission Deadline 2023.07.26 10:30 UTC+8): Swiss Round double competition system, score multiplied by 0.3 is included in the final total score
Final 2st Round (Submission Deadline 2023.08.09 10:30 UTC+8): Swiss Round double competition system, score multiplied by 0.7 is included in the final total score

Organizers

Haifeng Zhang

Associate Professor at Institute of Automation, Chinese Academy of Sciences

Nanjing Artificial Intelligence Research of IA

Email: haifeng.zhang@ia.ac.cn

Yan Song

Software Engineer at Institute of Automation, Chinese Academy of Sciences

Email: yan.song@ia.ac.cn

Yahui Cui

Software Engineer at Institute of Automation, Chinese Academy of Sciences

Email: yahui.cui@ia.ac.cn