---
title: "Discussion on \"Training a Self-Driving Car with PPO: What Worked, What Failed, and What I Learned\" | Hashnode"
description: "Discussion on \"Training a Self-Driving Car with PPO: What Worked, What Failed, and What I Learned\". Stable-Baselines3 PPO trained on CarRacing-v3 for 1M timestamps, how the +1-per-new-tile reward in the env creates a grass donut."
url: "https://hashnode.com/posts/training-a-self-driving-car-with-ppo-what-worked-what-failed-and-what-i-learned/695a4a569c469c1fea5f4666"
---

[Feed](https://hashnode.com/feed) /

Discussion

# [Training a Self-Driving Car with PPO: What Worked, What Failed, and What I Learned](https://practical-rl-self-driving-car-agent.hashnode.dev/training-a-self-driving-car-with-ppo-what-worked-what-failed-and-what-i-learned?utm_source=hashnode&utm_medium=feed)

Stable-Baselines3 PPO trained on CarRacing-v3 for 1M timestamps, how the +1-per-new-tile reward in the env creates a grass donut.

Jan 4 · 11 min read · practical-rl-self-driving-car-agent.hashnode.dev  
Read the article

-   [#self-driving-car-agent](https://hashnode.com/tag/self-driving-car-agent)
-   [#rl-agent](https://hashnode.com/tag/rl-agent)
-   [#car-racing-v3-environment](https://hashnode.com/tag/car-racing-v3-environment)
-   [#reinforcementlearning](https://hashnode.com/tag/reinforcementlearning)
-   [#reinforcement-learning](https://hashnode.com/tag/reinforcement-learning)
-   [#self-driving-cars](https://hashnode.com/tag/self-driving-cars)
-   [#ppo-reinforcement-learning](https://hashnode.com/tag/ppo-reinforcement-learning)

## Written by

[lakshmi prasanna](https://hashnode.com/@prasann)

@prasann

  
[View profile](https://hashnode.com/@prasann)

curiosity is the first step towards success

## Responses

Markdown supported · ⌘↵ to post

No responses yet.
