<!-- TITLE: Making rocket league robots --> <!-- SUBTITLE: --> I'll just paste some links here for now. I've installed the software, and hopefully I get the time / patience to play around with this. The competitive bar seems pretty low for now (e.g. glitchless, playing normally, not doing crazy wacky stuff) + [original source](https://www.reddit.com/r/RocketLeague/comments/6wcviq/how_to_create_a_rocket_league_bot/) + [latest version](https://www.rlbot.org/), which is actually awesome + [instructions on setting it up](https://www.youtube.com/watch?v=YJ69QZ-EX7k&feature=youtu.be) + [neural net-powered bot](https://github.com/SaltieRL/Saltie), which seems a bit infantile, but can do some things + [discord](https://discord.gg/syxUDt8) + [wiki](https://github.com/RLBot/RLBot/wiki) + [useful values](https://github.com/RLBot/RLBot/wiki/Useful-Game-Values) # The best in the game now + [flick](https://www.youtube.com/watch?v=VFRT16rsj9o), [air dribble](https://www.youtube.com/watch?v=VFRT16rsj9o) + [hoops!](https://www.youtube.com/watch?v=FGhgSEmObEU) - these are pretty good! + [tourney](https://www.youtube.com/watch?v=P9DG1MMIySc) - damn this sucks + [ReliefBot](https://github.com/tarehart/ReliefBot/tree/master/src/main/java/tarehart/rlbot) already has so much coded!! # Parsers for Rocket League replays This is probably necessary! At some point, training based on human behavior would make sense. At the most simple, I can copy motions directly, or at least know "if I hit it like X, the ball will go to Y", and trying to achieve that. + [list of parsers](https://github.com/rocket-league-replays/rocket-league-replays/wiki/Rocket-League-Replay-Parsers) + [ultimate replay list](https://ballchasing.com/) # The strategy I think I'll try to create little bots that know how to do something specific, and then patch them together later. I can also train a bot on X, and then train it on an extension of X (keep the ball in the air $\to$ pinch against the wall) # Creating a bot for aerialing 1. Figure out where the ball is going (reality) 2. Where do you want the ball to go? (wish) 3. How to make it do that... Could also just try some machine learning approach # Keep the ball on top of the car & steer? # Perfect pinch into the net