Thesis

Taking advantage of augmented reality technologies this thesis describes a solution to enhance the interaction between people and objects in museums. People are rarely able to manipulate objects in classical object-oriented exhibitions or, when it is possible, they do it with constrains. Museums can use augmented reality technologies more often in a playful way to enhance interaction and deliver new content among their public. This project was created for Faraday´s Museum (FM) at Instituto Superior Técnico in Lisbon. In order to increase the amount of interaction inside and outside Faraday´s Museum exhibitions we created and developed a gaming application that uses augmented reality technologies named Extended Play at FM. By using this application, the user/player can learn content about real objects allowing her/him to interact with digital replicas of the original pieces. Our goal is to provide interaction experiences in Museums questioning the relationship between users/players and objects to understand if people connect more deeply with the available content and learn through the process of interacting with digital augmented content. Starting with the main research question: Can augmented reality and human-machine interaction enhance and increase the player's engagement with artifacts available in museum spaces? The aim of this research is to provide interaction experiences in museums questioning the relationship between players and objects to better understand if people connect more deeply with the available content and learn through the process of interaction with increased digital content.

Nowadays, digital scenes, like games, tend to mostly use model animation, like facial expressions, and spoken dialogue, to convey emotion. However, those approaches are not always possible. This work explores the current approaches to text animation patterns, speech bubbles and background shapes and color, in order to convey emotion. It also aims to create a module that can control these aspects, and integrate it within a Virtual Tutoring application, which also utilizes model animation. We also present experiments, both comparing the existing base version, and final version of the application, and testing the emotional perception of each version, in order to understand if our approaches were correctly implemented. It was possible to observe that the addition of such details led the system to be recognized as better animated, when compared with the earlier version, however it when tested alone, some emotions ended up being less recognized, while others had a better impact on the application. In the end, we observe that such details can help improve the emotional recognition of a scene, however the additional information on one channel can create noise and bias on the other emotional channels, leading to worse results in certain occasions. Nonetheless, they can and should be used in order to reinforce emotional recognition, without creating contrast with the leading emotional channel.

This document describes the development of a new online competition platform to support the future of the Geometry Friends Game AI competition, an artificial intelligence competition based on the Geometry Friends game. The Geometry Friends game is a cooperative 2D platformer whose main characteristics are visited in the beginning of this work, along with how the competition has been running until now. Several other AI competition platforms are studied, as well the practices which are considered to be state of the art to run good and prosperous game based AI competitions. After analysing how the current Geometry Friends competition fairs against the state of the art, several requirements that the new platform should meet were laid out, which included facilitating competition participation online and automating several submission handling processes. The new platform was developed mostly from scratch, consisting of a new website, which both participants and competition organizers may interact with, and a background program in charge of fully handling the received submissions, and more. Virtualization was used to create secure, fair and reusable evaluation environments. Different features of the new platform were tested by several people across three different test scenarios. The new website was concluded to have an above average usability, while the submission handling program worked as expected throughout the testing phase. It was also possible to conclude that the main objective was accomplished, since a new platform was successfully developed.

Procedural content generation is a big part of the gaming industry nowadays, from generating levels, to weapons and even characters, this method of content creation has multiple uses and covers a variety of areas. In this work, we tackle the existing gap in procedural content generation, specifically for cooperative games by presenting a solution of our own. Our work is comprised by a level editor with the capability of generating cooperative levels for the game Geometry Friends. The level generator was based of the algorithm Monte Carlo Tree Search and a flood field based evaluation algorithm. We conducted experiments to see if we were successful in developing such tool and if the levels generated were interesting, fun and were in fact cooperative. Our results suggest that we achieved our goals in creating a tool that produced cooperative levels that are engaging.