Mario AI

Mario AI 是强化学习的基准环境。与最初的任天堂版本一样，Mario AI 中的游戏玩法包括将受控角色（即马里奥）移动到横向观看的二维关卡中。马里奥可以左右行走、奔跑、跳跃和（取决于他所处的状态）射击火球。重力作用在马里奥身上，因此必须跳过悬崖才能越过它们。马里奥可以处于以下三种状态之一：小、大（可以通过跳到敌人身上杀死敌人）和火（可以发射火球）。

Mario AI is a benchmark environment for reinforcement learning. Similar to the original Nintendo version, the gameplay of Mario AI involves moving the controlled character (i.e., Mario) through side-scrolling 2D levels. Mario can walk left and right, run, jump, and (depending on his current state) shoot fireballs. Gravity acts on Mario, so he must jump over cliffs to traverse them. Mario can be in one of three states: small, big (which allows him to defeat enemies by jumping on them), and fire (which enables him to launch fireballs).

搜集汇总

背景与挑战

背景概述

Mario AI是一个用于强化学习的基准环境，基于超级任天堂游戏，模拟马里奥在二维关卡中的移动、跳跃和攻击等动作。该数据集由哥本哈根大学等机构于2009年发布，旨在支持相关研究。