Thesis

Different game technologies as the Microsoft Kinect and the Nintendo Wii Remote has been remarkable for the growth of motion gaming. Even so, competition between players seems to be integrated in most video games in the video game industry. Yet, collaboration features can carry new game experiences that can interest some players. This thesis discusses some distinct Kinect gestural interaction and relates them to if they can enhance the collaboration feeling between players. The controls chosen for each interaction are simple and were tested in a casual collaboration game called Geometry Friends. From the different controls, four different types of interaction were chosen to be tested: Hand interaction, Full-Body interaction, Tangible interaction and Collaboration gestures interaction. Furthermore, a Wii motion control interaction was properly compared within many different aspects. Natural gestures, collaboration sense, gesture recognition, gesture exhaustion and space issues are some of the aspects compared between all interaction tested. The tests revealed that Geometry Friends can achieve true collaboration sense and this sense vary with an interaction change, yet no direct correlation was found between natural/playability factors and the collaboration sense of the players. All this contributed to conclude that more tutorials and difficulty changes should be evaluated carefully when changing the game interaction, especially when changing to a gestural interaction.

The objective of this report is to show the results of our approach of a motivational drive-based architectures using a physics engine for modulating the agent's action selection processes. Rigid-bodies are used to represent abstract mind concepts, like needs, actions and perceptions, and the collision between these objects triggers agent behaviour. This thesis presents our conceptual model that integrates the standard motivational drives model into a physical context, along with our implementation, tests and results of that same model. We have also researched how different architectures that use motivational drives implement concepts like, perception and action selection, memory and learning, emotions and personality, in order to understand key feature that our system must have in order to produce believable behaviour.

Nowadays, video-games have reached unprecedented bounds in terms of quality, realism, immersion and gameplay. A deciding factor in this conjecture has been the implementation of more intricate, intelligent behaviour on enemies and allies Non-Playable Characters(NPCs), in order to provide superior wit and challenge to human players. Our task is to apply squad intelligent behaviour in order to maximize the efficiency of Micromanaging units in a Real-Time Strategy (RTS) game. After describing Micromanaging and RTS games, using Starcraft as our main reference, attempts made in the 2010 AIIDE Starcraft competition was analysed, putting them against each other and a human player, to identify possible faults. Afterwards, research was made focusing on AI Squad intelligence techniques not only found in RTS games, but also other genres that feature such mechanics like First-Person Shooters, to find possible ways of addressing the problem and compile them into a Starcraft artificial player architecture with the objective of competing against human players and other micromanagement artificial players. The developed artificial player proved to be a challenge against casual and seasoned players alike, and had a solid performance against another selected artificial player.

Games have become a powerful tool to help with learning and training, but they sometimes lack the believability necessary for a good training simulation, especially if that simulation deals with human behaviours. Our work aims at creating a simulation for team management training. For this simulation, agents will form a team and will not be focused only in the task. Their personality and motivational state will influence the resolution of the task. This document will present the initial research done in trying to figure out the initial problem and the following research originated from the defined scope of the work. The findings were then compiled into a conceptual model which was implemented as explained in the architecture. We also explain how the game gives life to the proposed conceptual model. We hope to prove that this work is relevant with the tests done and propose some future improvements that can be done using our work.