UOttaHack 8 Best Hardware Hack - GURT (Ground Unit Response Tank)

Devpost | gurt.tech (Robot is currently decommissioned so website will not work D: rest in piece gurt)

Features

• Real-time Video Feed: Low latency streaming from Pi using a custom-built QNX camera driver (optimized from ~3fps to being able to run consistent 30fps).
• Low Latency Control: WebSocket-based control loop.
• Hardware Integration: Real-world hardware control with Raspberry Pi 4 and Arduino.
• Two Game Modes:
  • Casual: Free to play.
  • Ranked: Pay 0.1 SOL Entry Fee -> Win -> Get 0.18 SOL Payout.
• Leaderboard: Track top pilots.

This project implements a real-time remote control system with video streaming using WebSockets. It consists of a central FastAPI server, a web-based client for control, and a hardware client running on a Raspberry Pi 4.

Hardware Architecture

Our system uses a dual-controller architecture for reliability and performance:

1. Raspberry Pi 4 (Main Controller):
   • OS: QNX 8 (Unix-based RTOS).
   • Responsibilities: Runs the Python client (pi_client.py), handles WebSocket communication, and streams video.
   • Custom Camera Driver: To overcome QNX's default limitations, we developed a native camera driver that boosted frame rates from a stuttering 3 FPS to a smooth 30 FPS. Code adapted from camera_example3_viewfinder
2. Arduino (Motor Controller):
   • Responsibilities: Receives serial commands from the Pi to control the physical motors of the robot.
   • Logic: Located in turning.ino.

Prerequisites

• Python 3.8+
• Hardware setup (optional, simulation supported):
  • Raspberry Pi 4 with QNX 8
  • Arduino (for motor control)
  • Webcam / Camera Module

Setup

1. Clone the repository:

   git clone <repository_url>
   cd uottahack-8

2. Create and activate a virtual environment:

   # Windows
   python -m venv venv
   .\venv\Scripts\activate
   
   # macOS/Linux
   python3 -m venv venv
   source venv/bin/activate

3. Install dependencies:

   pip install -r requirements.txt

4. Solana Devnet Setup: The server acts as the "House" and needs a wallet to hold entry fees and pay out winnings.

   • Start the server (step below).
   • Look for the [SOLANA] House Wallet Public Key in the terminal output.
   • Send Devnet SOL to this address so it has funds to pay out winners!
   • You can get Devnet SOL from faucet.solana.com.

Running the Project

1. Start the Server (Cloud/Laptop)

This starts the central communication hub.

python server.py

Open your browser and navigate to http://localhost:8000.

2. Start the Pi Client (On QNX 8 / Raspberry Pi)

This runs on the Raspberry Pi 4. It connects to the server, receives control commands, and streams video using our custom QNX driver.

python pi_client.py

We have implemented 2 camera methods. The primary method opens a camera using OpenCV, which will works on most OSes (Windows, MacOS) as well as Linux based OSes. For QNX, we have developed our own camera driver as a secondary method. This requires the camera_example3_viewfinder binary which we have precompiled in this repository.

Note: You can also run this on your laptop for testing; it will fallback to your webcam.

Usage

1. Ensure both the Server and Pi Client are running.
2. Open the web interface (http://localhost:8000).
3. The web interface sends control data to the server.
4. The server relays these controls to the pi_client.py.
5. The pi_client.py forwards motor controls to the Arduino and streams video back.

Troubleshooting

• Port 8000 already in use: Edit server.py and change the port in uvicorn.run(..., port=8000).
• No Webcam: The pi_client.py will automatically fall back to generating visual noise if no webcam is detected.
• Connection Refused: Ensure the server is running before starting the client.