DuckFightClub

Product Introduction

1. Definition: DuckFightClub is a structured, livestreamed competitive robotics event and platform centered on the MicroDuck, an open-source bipedal robot. It is a simulator-first robot sumo league where participants train and submit reinforcement learning (RL) policies, not physical hardware.
2. Core Value Proposition: It exists to create an accessible, high-stakes, and entertaining competitive arena for robotics and AI developers. It solves the problem of high cost and fragility in physical robot combat by shifting the competition to a simulated environment, allowing teams to focus on AI policy development, machine learning training, and strategic gameplay for a global audience.

Main Features

1. The QuackDown Tournament: A single-elimination, eight-team bracket where AI policies for the MicroDuck robot compete in a best-of-three simulator sumo format. The event features a structured schedule from team registration and policy training to a final livestreamed championship, culminating in the awarding of the Golden Beak Belt to the winning team's policy.
2. MicroDuck Simulator & RL Framework: The competition is based on Pollen Robotics' open-source MicroDuck simulator and reinforcement learning repository. This provides a standardized, reproducible physics environment where teams can develop, train, and test their control policies for pushing, dodging, and ring awareness without needing physical robot hardware.
3. Virtual Sponsorship & Broadcast Integration: The platform offers monetization through integrated virtual arena sponsorships. Brands can purchase digital placements on the mat center, apron boards, or as a naming partner on the back-wall marquee, which are featured during the professional-style livestream broadcast on YouTube.
4. Local Chapter & Showdown Hosting: DuckFightClub provides a format for third-party organizers to host local events ("showdowns") in their city or at conferences. This decentralized model helps grow the community and allows for regional competitions, like the confirmed "Prise de bec" event in Paris.

Problems Solved

1. Pain Point: High barrier to entry for competitive robotics. Traditional robot combat requires significant investment in durable hardware, mechanical engineering, and repair, which limits participation. DuckFightClub eliminates hardware costs and damage by making the core competition purely software-based in simulation.
2. Target Audience: The primary personas are reinforcement learning engineers, robotics researchers, university AI/robotics clubs, and open-source software developers interested in applied ML. Secondary audiences include tech event organizers looking for engaging content and brands seeking to sponsor innovative AI/robotics communities.
3. Use Cases: Essential for teams wanting to benchmark their RL algorithms in a dynamic, adversarial environment; for educators seeking a compelling project for robotics courses; for companies like Pollen Robotics to drive adoption and testing of their MicroDuck platform; and for content creators producing live esports-style engineering entertainment.

Unique Advantages

1. Differentiation: Unlike traditional robot combat (BattleBots) or simulated game AI competitions (like those for StarCraft or Dota), DuckFightClub uniquely combines the spectacle of sports entertainment with the technical challenge of training RL policies for a specific, real-world open-source robot platform, all within a controlled simulator.
2. Key Innovation: The product's core innovation is its "policy-as-fighter" paradigm. The competitive unit is not a team's custom robot design, but their trained AI control policy running on a standardized robot model. This shifts the competition entirely to software intelligence and training strategy, ensuring a level playing field focused on algorithmic prowess.

Frequently Asked Questions (FAQ)

1. Is DuckFightClub using real robots? No, the QuackDown championship is conducted entirely within the MicroDuck simulator. This allows for intense, damage-free competition and lowers the barrier to entry, as teams compete with AI policies instead of physical hardware. Future events may incorporate prototypes if available.
2. How do you train a DuckFightClub policy? Teams use the open-source MicroDuck reinforcement learning (RL) framework and simulator from Pollen Robotics. They write a control policy and use RL techniques to train it through millions of simulated sumo scenarios, optimizing for behaviors like ring control, opponent pushing, and recovery from near-falls.
3. What are the rules for the simulator sumo fights? Fights are best-of-three rounds within a virtual circle. A round is won by forcing the opponent's policy out of the circle. All policies must use the approved simulator control interface and are frozen before the event to ensure fair and reproducible matches during the livestream.
4. Can anyone sponsor DuckFightClub? Yes, the event offers virtual sponsorship packages. Brands can sponsor the mat center, individual apron boards around the ring, or become the naming partner with a marquee on the virtual arena's back wall. These digital assets are integrated into the live broadcast graphics.
5. How does the team registration and bracket work? Eight teams are selected for the main QuackDown bracket. The tournament is single-elimination with no byes, starting with four quarterfinal matches. The schedule includes a registration period, policy submission freeze, bracket reveal, and a final livestreamed event to crown the champion.