A Spatial Lexical Analyzer and 3D Grammars that Recognize Voxel Based Structures Using Linear Positional Grammars in Minecraft

• Geraldo Xexéo UFRJ
• Leonardo Santos UFRJ

Abstract

We introduce the use of 3D grammars for the identification of objects formed by voxels, i.e., discrete element cells that compose a 3D grid. We also show how to apply these grammars in a bot that can recognize structures in Minecraft. Our main contribution is a new spatial lexical analyzer that can decide the direction it must follow to obtain the next input by looking at the future positions and the expected next token in a Linear Positional Grammar. Moreover, we define different positional relations and specific grammars for various Minecraft structures.

Keywords: Video games, Three-dimensional displays, Neural networks, Games, Syntactics, Chatbots, Search problems, 3D Grammars, Voxel, Object Recognition