https://github.com/rlbot/rlbottraining

Customizable training for Rocket Leage bots of any language.
https://github.com/rlbot/rlbottraining

Last synced: 7 days ago
JSON representation

Customizable training for Rocket Leage bots of any language.

• Host: GitHub
• URL: https://github.com/rlbot/rlbottraining
• Owner: RLBot
• License: mit
• Created: 2018-12-24T13:04:13.000Z (almost 6 years ago)
• Default Branch: master
• Last Pushed: 2021-07-12T10:42:57.000Z (over 3 years ago)
• Last Synced: 2024-10-07T02:36:19.129Z (about 1 month ago)
• Language: Python
• Homepage:
• Size: 449 KB
• Stars: 7
• Watchers: 9
• Forks: 4
• Open Issues: 3
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# RLBotTraining

Customizable training for Rocket League bots of any language.

## Tutorial: [Click Here](https://youtu.be/uGFmOZCpel8?list=PL6LKXu1RlPdxh9vxmG1y2sghQwK47_gCH)

## Installation

 - Install Python 3.7 or later

 - `pip install rlbottraining`

## Features at a glance

 - Uses `RLBot` config files to support bots in any programming language

 - Reproducable training setups with seeded randomness

 - Customizable Pass/Fail criteria

 - Importing of shots/playlists from BakkesMod training

 - Playlists of exercises

 - Automatic reloading of both bot and exercise code

 - Imports from bakkesmod training

## Architecture

 - At the entry point (`run_exercises.py`) we decide which `GraderExercise` should be run and the `config_path`s define the RLBot setup to be used (which bots, which mode).

 - Each `GraderExercise` composes together the initial game state and how the bot is judged. The responsibility of judging the bot is handled by `Grader`s

 - A `Grader` decides whether to continue the exercise or to `Pass`/`Fail` the exercise by looking at each tick. A `Grader` may optionally gather metrics (e.g. "time until goal") to help track bot performance as the bot is improved.

 - Separation of responsiblity: This repository is designed to make it nice to define new exercises whereas the training API of the `RLBot` framework provides the minimum features for any training to occur.

## Tips for writing your own exercises:

 - Always subclass `GraderExercise` rather than `Exercise`. Using `Grader`s will allows you to share termination conditions, safely store state across ticks and provide metrics.

 - Compose your `Grader`s. For instance, by using `CompoundGrader`

 - Provide meaningful error messages in your `Grader`s by subclassing `Fail`

## Future direction

 - Provide an example of striker vs goalie

 - Put metrics (`Grader.get_metrics()`) into a database

 - A way of browsing past runs, including metrics

 - Reproducing failed exercises

 - Visualization of metrics

 - Continous integration of different bots