Simulator-based matches
All bouts run in the MicroDuck simulator instead of using physical robot hardware, allowing policies to compete in a controlled virtual environment.
Simulator sumo for reinforcement-learning teams
DuckFightClub is a livestreamed simulator-sumo event for reinforcement-learning teams. Participants submit MicroDuck control policies, then compete in a fixed virtual ring for the Golden Beak Belt rather than operating physical robots.
The event is structured as an eight-team, single-elimination tournament. Teams train movement, pushing, dodging, ring awareness, and recovery in simulation, lock their final policies before the broadcast, and advance through best-of-three bouts based on whether their policy can keep its duck inside the circle while forcing its opponent out.
Competition and simulator features
Reinforcement-learning policies
Teams create controllers for movement, balance, pushing, feints, and recovery, with reinforcement learning supported for training behaviors in simulation.
Standardized ring setup
The event uses a fixed circular ring with the same spawn spots and physics for every match, supporting consistent match setup.
Best-of-three bouts
A round is won when one policy drives, trips, or pushes the opposing policy outside the circle. Bouts are best of three rounds.
Eight-team bracket
Eight teams enter a single-elimination bracket with four quarterfinals, two semifinals, and one final; there are no first-round byes.
Pre-event policy lock
Final policies are submitted before the event and must run at the event settings through the approved control interface.
How DuckFightClub can be used
Benchmark control policies
RL teams can develop and compare controllers for balance, edge awareness, pushing, dodging, and recovery under a shared event configuration.
Run pre-event scrimmages
Teams can use public simulator scrimmages scheduled before the event to test movement and ring behavior before the final policy submission.
Showcase robotics research
The broadcast format gives technical communities a way to present policy behavior through bracket matches, replays, and on-screen match graphics.
Sponsor a virtual arena
Sponsors can purchase placements on the apron boards, mat center, or back-wall marquee, with additional naming, bout, and stream partnership options described by the site.
Host a local showdown
Cities, conferences, and labs can propose local DuckFightClub showdowns using the format for their own events.
Pros and Cons
Pros
- Provides a defined eight-team tournament format with no first-round byes.
- Uses consistent ring geometry, spawn positions, and physics for match reproducibility.
- Supports reinforcement-learning workflows focused on balance, pushing, dodging, and recovery.
- Includes a public broadcast format with bracket coverage, replays, overlays, and sponsor placements.
- Awards the winning team the Golden Beak Belt after the final.
Cons
- The YouTube channel and some participation details are not yet fully specified on the source site.
- Entries must run at event settings through the approved control interface, which may limit policy portability.
- The competition is simulator-based; physical MicroDuck hardware is not used for the scheduled bouts.
FAQ
What is DuckFightClub?
DuckFightClub is a livestreamed simulator competition in which MicroDuck control policies compete in a robot-sumo-style bracket for the Golden Beak Belt.
Are physical robots used in the matches?
No. The bouts run in the MicroDuck simulator rather than with physical robots. The source site says MicroDuck hardware is not yet released.
How do teams train their entries?
Teams write control policies for movement, balance, pushing, feints, and recovery, and may use reinforcement learning to train behaviors such as ring control, dodging, and edge recovery in simulation.
How does the competition format work?
Eight teams compete in a single-elimination bracket. There are four quarterfinals, two semifinals, and a final, with no byes. Each bout is best of three rounds, and the first policy to win two rounds advances.
Will the event be livestreamed?
Yes. The event is scheduled to stream live on YouTube, although the channel is listed as to be announced. The broadcast is described as including match cards, replays, overlays, and sponsor placements.
Quick Facts
- Category
- Reinforcement-learning simulator competition
- Platform
- MicroDuck simulator
- Match format
- Eight-team single elimination; best of three rounds
- Scheduled event
- QuackDown on October 2, 2026
- Broadcast
- Live on YouTube; channel to be announced
- Source domain
- duckfight.club
DuckFightClub 대안
GameGrad
GameGrad turns PDF study material into quiz questions that appear inside arcade games, so users can practice active recall while they play. It offers a free tier with one PDF and a Pro plan with more uploads, more questions, and answer explanations.
DailyPlay.ai
DailyPlay.ai is a behavioral growth platform that uses AI site scanning and micro-games to help brands improve engagement and conversion. It supports web, mobile, and other digital touchpoints, with a one-line script setup and a free engagement scan.
ChessWoodie
ChessWoodie is a chess tactics training app based on the Woodpecker Method. It helps players build pattern recognition through repeated puzzle cycles, track speed and accuracy, and review missed tactics from their own games.
PixelForge
PixelForge turns a single photo into a game-ready 2D character sprite pack, including a 4×4 sheet, transparent frames, direction strips, and looping walk GIFs. It is designed for engine-ready use in Godot, Unity, Phaser, GameMaker, and web games.
Loniva
Loniva is a mobile puzzle app that offers a calm daily mix of games across memory, attention, language, speed, logic, and focus. It runs on iPhone, iPad, and Android, with a free 7-day trial and a free daily workout after the trial.
Mixels.ai
Mixels.ai is a browser-based AI pixel art generator for creating 8-bit characters, sprites, avatars, and photo-to-pixel conversions. It is designed for users who want grid-aligned retro artwork without installing software.