Reconnaissance Blind Chess

Join the RBC Research Community


Game Overview

Reconnaissance Blind Chess (RBC) is a chess variant designed for new research in artificial intelligence (AI). RBC includes imperfect information, long-term strategy, explicit observations, and almost no common knowledge. These features appear in real-world scenarios, and challenge even state of the art algorithms.

Each player of RBC controls traditional chess pieces, but cannot directly see the locations of her opponent's pieces. Rather, she learns partial information each turn by privately sensing a chosen 3x3 area of the board.

RBC's foundation in traditional chess makes it familiar and entertaining to human players, too!

Join the study and fun now by reading the rules, building a bot, or playing a game.


NeurIPS 2021 Competition

We are hosting another competition as part of NeurIPS 2021! Can you compete with students, scientists, and engineers from all over in part of the world's largest AI conference? Use the links to the left for more information.


Hidden Information Games Competition

DeepMind and the Czech Technical University in Prague are hosting the Hidden Information Games Competition (HIGC) to test AI bots in large two-player zero-sum games with imperfect information. They have selected Reconnaissance Blind Chess as one of three games in their competition. Can your algorithm compete with others in the HIGC too?


2020 Leaderboard Challenge

Congratulations to Gregory Clark, author of penumbra, and Kyle Blowitski and Timothy (T.J.) Highley, authors of La-Q Bot, who won $1000 and $500 respectively in our most recent scheduled online competition! Hear more in an interview with JHU/APL and Clark. Thank you to all who participated.

Although the cash prize has been awarded, the competition to create one of the top bots on the leaderboard is ongoing!


NeurIPS 2019 Tournament

Congratulations to all the participants of our NeurIPS 2019 tournament! Gino Perrotta and Robert Perrotta won first prize with StrangeFish, which had an overall record of 464-40 and a winning record against every other bot. Timothy (T.J.) Highley, Brendan Funk, and Laureen Okin won second with LaSalle Bot, which had a record of 444-60. In addition to replays of all tournament games, more information is available in our PMLR paper and the NeurIPS videos.


Current Bot Leaderboard

Bot and human player ratings will appear on this board after completing at least 10 ranked games. See the Ranking Protocol page for more information.

Rank User Rating Num Ranked Matches
- - - -
- - - -
- - - -
- - - -
- - - -
- - - -
- - - -
- - - -
- - - -
- - - -
1