yolo-brawlers

KO | EN

gitlite — search

#typescript #ai-agents #deepseek-harness #dsh-plugin #open-source #ai #cli #dsh #claude-code #codex #developer-tools #react

★ 77 Open GitHub ↗

Use cutting-edge computer vision to turn into a real-life robo-fighter! With only a laptop, control a tabletop boxing mech, dodging enemy blows and unleashing devastating attacks! Get ready to rumble!

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Yolo Brawlers

Yolo Brawlers is an OpenCV project created for HackED 2025! You can find more information about this project
on [DevPost](https://devpost.com/software/yolo-brawlers) regarding the project development process, objectives, and motivation.

## Description

Yolo Brawlers is an interactive real-time motion-controlled game where two players control robotic fighters using their own body movements. Using [YOLO V11 Pose Detection](https://github.com/ultralytics/ultralytics), the system translates human fighting moves into commands for the robots, making them punch, weave, and guard dynamically.

## Gallery
![IMG_4483](https://github.com/user-attachments/assets/f2ba38a2-d61a-440a-8a33-d0f62f6f3232)

![19C21752-C7D1-428F-B070-E98F84A3DB98_4_5005_c](https://github.com/user-attachments/assets/cdbb7106-7beb-42ac-8e96-8922efde17bd)

![image](https://github.com/user-attachments/assets/fb3f225c-a0a8-469a-aaa4-b4840ad9dbf2)

![image](https://github.com/user-attachments/assets/d8341888-6c11-46dc-a18b-117827af595b)

## How It Works

1. Pose Detection (YOLO + OpenCV): The system uses YOLO pose estimation to track players' movements in real time.
2. Move Classification: The AI recognizes different fighting poses (punches, weaves, guard stance).
3. ESP32 Controller: The firmware uploaded to the ESP32 controls servo motors connected to the ESP32 GPIO pins
4. Robot Control: Detected player actions are sent to the ESP32 controller using sockets at its WiFi access point. The data is used to control the robots to mimic boxing maneuvers like punching, weaving and guarding.
5. Multiplayer Battle: Two players compete by controlling their own robotic fighters with body movements.

## Tech/Hardware Stack

-   Python, PyQt5 → Graphical user interface (UI) for starting/stopping the game.
-   YOLO (Ultralytics) → Pose detection for movement tracking.
-   OpenCV → Real-time camera feed processing.
-   C++, ESP32, Servo Motors → Robot motion control.

## Getting Started

1. Install conda on your local machine through this [link](https://docs.anaconda.com/anaconda/install/) or through the OS package manager provided.
2. For Apple Silicon based chips:
    - Create your conda environment using `conda create -n ultralytics-env python=3.12 -y`.
3. For AMD/Intel based chips:
    - Create your conda enviroment using `CONDA_SUBDIR=osx-arm64 conda create -n ultralytics-env python=3.12 -y`.
4. Enter the conda environment by entering `conda activate ultralytics-env`.
5. Install necessary python libraries in conda environment using `pip install -r requirements.txt`.
6. Run the project inside the UI folder using `python3 finalUI.py`.

# Running the Program
1. Ensure you have the firmware in the /controller PlatformIO project installed on the ESP32 Microntroller
2. Power the Microcontrollers and on your laptop, connect to their WiFi network (ESP32_Servo_Control_Red or ESP32_Servo_Control_Blue)
2. Navigate to the ./UI directory and run `python finalUI.py`
3. Follow the instructions on the UI

## Notes

-   For Macbook Intel Chip you need to install `pip install "numpy<2"` manually.
-   For any Apple Silicon Chips you cannot access the keyboard controller.

## Resources

-   [Ultralytics Coco Pose Estimation](https://docs.ultralytics.com/datasets/pose/coco/)
-   [Ultralytics Pose Estimation Tasks](https://docs.ultralytics.com/tasks/pose/)

## Team Members

1. [Aron Gu](https://github.com/arongu321)
2. [Sami Jagirdar](https://github.com/Sami-Jagirdar)
3. [Jared Drueco](https://github.com/jdrco)
4. [Antonio Martin-Ozimek](https://github.com/antonio2uofa)

yolo-brawlers

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README