Adaptive Generation of Characters for Tabletop Role Playing Games

Abstract

This paper tries to solve the problem where automated character creation for Tabletop RPGs is completely random and does not fully pleases certain players, especially those that have difficulty in being creative and adapting characters to their personal tastes. To accomplish that, the usage of a Evolutionary Algorithm (EA) was proposed, in conjunction with the user's input (positive or negative) during generations, to create characters focused on having traits that are catered to each individual's taste. The generated traits are 'Race', 'Class', 'Size', 'Visual Trait', 'Personality' and 'Quirk'. The fitness calculation for each individual was made using trait tables with weights previously defined by the programmer, and posteriorly by the user's evaluation of the characters during execution. The project was made using the Unity game engine with C# language, to take advantage of the tool's easy interface creation. It was produced as a phone application, for ease of distribution and general use. The application was tested with a variety of users, and the results, presented in the form of graphs, indicate that the rejection rate was almost null, with the great majority of users recognizing the gradual improvement of suggestions between generations, and evaluating positively the characters generated by the algorithm.