RL-Racecar 🏎️

Goal 🥇

Use reinforcement learning to find the fastest way around a track.

Tools 🧰

• Python
• NEAT
• Pygame

---

Challenges 📈

• I first started by trying to implement my own Q-Learning algorithm, but I did not get the results I was looking for.
• I had never used NEAT before, and learning how to use it took a bit of time.
• Pygame surfaces also took a while to get used to. I started by making the car just a surface, but ended up with it as a sprite.

---

Results 🌟

• All goals were accomplished and the finished product can be used!

---

Directions 📋

• Ensure that you have installed git and Python on your device.

• Open the terminal or GitHub Desktop and paste this command https://github.com/Ryan-W31/RL-Racecar.git

• Once the repo has been cloned, enter into the main project directory from the terminal.

• Create a virtual python envirnment using either Anaconda or virtualenv.

  virtualenv: 1. pip install virtualenv 2.venv {envirnoment name}
  Anaconda: 1. Install Anaconda 2.conda create {envirnment name}

• Using pip, download the required packages using the requirements.txt

• From the main project directory, paste this command in the terminal: python3 main.py ...If you want to see the radars and checkpoints, use this command python3 main.py --debug

• Enjoy!

---

Resources Used 🧠

• [https://github.com/juangallostra/procedural-tracks]
• [https://github.com/DEV-D-GR8/Self-Driving-Car]
• [https://github.com/maxontech/DriveAI/tree/master]
• [https://www.youtube.com/watch?v=r428O_CMcpI]