· · · Institutional Repository

To increase a player's immersion in the game world, its objects should behave as one would reasonably expect. For this, it is now becoming increasingly clear that what game objects really miss is richer semantics, not eye-catching visuals. Current games' lack of semantics is mostly due to the complexity of designing semantic objects. If game designers would have to design all semantics by themselves, and game programmers would have to implement the handling of these semantics, game development time would increase enormously. The goal of this research is to improve the semantics of game objects (or more properly, entities) in virtual worlds, resulting in more and better object interaction in games. This thesis proposes a solution in the form of services, describing interaction possibilities between entities, and a structure that makes it easier to specify them. An example of this is the service of a vending machine, which exchanges a coin supplied by a player for a soda. By introducing several components, such as classes, attributes and actions, a service is defined as the capacity of an entity to perform a particular action, possibly subject to some requirements. To incrementally specify and add services to game objects, a three-phased methodology is presented. This approach has been implemented and validated by means of a prototype system, which enables a simple and intuitive definition of services in an integrated environment. By using two different editors, services can be defined during the game development process. In turn, a semantics engine is charged with all service handling during a game itself. It is concluded that if game designers are presented with the tools to easily add semantics to game objects, these objects become aware of their services, therefore facilitating more and better object interaction, resulting in a much deeper gameplay experience than in current games.

In the past few years, traffic incident management has been a subject of interest in both politics and research. After an incident has occurred, incident management measures are intended to clear the motorway for traffic, inform other road users and treat victims as soon as possible. These measures largely constitute formal agreements, procedures and coordination efforts between all parties involved. Although the new incident management measures currently in effect in the Netherlands have improved the time required to clear an incident scene, information technology is expected to decrease the required time even further by improving communication and situational awareness and supporting analysis and evaluation of the incident response. In this scope, a new IT system is developed, called the Incident Management Information and Communication System (IMICS). Based on an existing framework, the Java Agent DEvelopment framework (JADE), IMICS offers a blackboard like functionality through which the police, fire brigade, ambulance, traffic control centres, the shared control rooms and the Department of Public Works can update and share incident and task information both at the control rooms and at the incident scene. In order to provide clear and unambiguous information, an ontology and an information management system have been designed that closely follow the incident management procedures. Finally, a part of the IMICS design has been implemented and tested in a lab setting as a proof of concept. The results show IMICS improves availability and reliability of data in a lab setting.

Traffic congestion has negative impact in our life. In 2001, the European Commission estimated costs of congestion at around 120 billion Euro. The research of Intelligent Transportation System (ITS) becomes interesting since infrastructure expansion in not always feasible. In order to find out the impact of automated vehicle and intelligent infrastructure to traffic performance, we develop a model of hierarchical vehicle controller and intelligent infrastructure and implement them in a simulated environment using Multi Agent System concept. Several traffic scenarios are deployed to observe the effect of the designed vehicle controller and the model of intelligent infrastructure in traffic.

Electron microscopy makes it possible to magnify samples at almost atom level by using electron beams to scan the sample. These images could be improved in terms of noise and sharpness. In this research we looked into the possibilities to use post-processing filters for improving the quality of electron microscopic images.

This thesis describes our internship at Tenuki. The assigment was to impress visitors of Tenuki with the traffic information of Tenuki.

This thesis describes how an automatic lip reader was realized. Visual speech recognition is a precondition for more robust speech recognition in general. The development of the software comprised the following steps: gathering of training data, extracting meaningful features from the obtained video material, training the speech recognizer and finally evaluating the resulting product. First, research was done to gain insight on the theoretical aspects of automatic lip reading, the state of the art, speech corpus development, face tracking and feature extraction. Gathering training data came down to the recording and composing of a new audio-visual speech corpus for Dutch. With frontal and side images of 70 different speakers recorded at a frame rate of 100 frames per second this is the most diverse corpus currently in existence. Analysis of the new data corpus shows an increase in quality compared to other corpora. Visual information is obtained by searching the video footage. Using Active Appearance Models, points of an a priori defined model of the lower half of the face are tracked over time. Based on the model point coordinates, distance and area, features are computed that are used as input to the speech recognizer. Training was accomplished by presenting labeled training data to viseme-based Hidden Markov Models that model speech production. In a few steps the model parameters were adjusted, so that it could be used to perform recognition of visual speech signals from then on. The recognizer was implemented using tools from the Hidden Markov Model Toolkit. The results of a visual speech recognizer based on training data from a single person depend on the utterance type of the unlabeled data. For the simple word-level task of digit recognition 78% was recognized correctly with a word recognition rate of 68%. For letter recognition tasks it did not perform nearly as well, but considering the limitations that the use of visemes over phonemes imposes, these results are at the expected level. The data corpus and visual speech recognizer will be a valuable asset to future research.

At the moment current anticoagulation dosing computer programs use of relatively simple algorithms and provide a useful dose proposal in about half the cases. In the other cases medical staff has to come up with a dose and a return date. This thesis investigates the possibilities of supporting medical staff in anticoagulant dosing by extending current computer programs. Research has been done for improving the current algorithms and incorporating expert knowledge in a computer system. A literature survey showed that a lot of effort already has been made in improving algorithms. The ICAD seems to be the most promising, but still fails in about 20% of the cases to provide a dose or return date, mostly due to special circumstances. My aim was to find out whether the medical expert's knowledge was retrievable and if it was suitable for developing an expert system. Therefore I interviewed dosage doctors and dosage advisors in two Thrombosis Service Departments. The majority of the interviews were held at Star-MDC in Rotterdam, where I work as a Laboratory Manager. A couple of interviews were also held at Saltro, a comparable organization in Utrecht. Knowledge elicitation and representation techniques were applied and a prototype was developed. Further interviews were held, based on the prototype and a OODA-Loop technique. A lot of existing protocols at the Thrombosis Service organizations seemed to be suitable for converting in IF-THEN rules, as a basis for a rule-based expert system. Also the reasoning process of the relative low complex cases are obtainable in an expert system. This is specifically interesting, because the low complex cases are the majority of cases that have to be reviewed by medical staff.

New ideas to improve automatic speech recognition have been proposed that make use of context user information such as gender, age and dialect. To incorporate this information into a speech recognition system a new framework is being developed at the MMI department of the EWI faculty at the Delft University of Technology. This toolkit is called Gaia and makes use of Dynamic Bayesian networks. In this thesis a basic speech recognition system was built using Gaia to test if speech recognition is possible using Gaia and DBNs. DBN models were designed for the acoustic model, language model and training part of the speech recognizer. Experiments using a small data set proved that speech recognition is possible using Gaia. Other results showed that training using Gaia is not working yet. This issue needs to be addressed in the future and also the speed of the toolkit.

A study is presented of the cover song retrieval problem, in which multiple performances of the same musical work are sought. The problem is considered in the context of content-based audio retrieval. In order to gain more insight into the current state-of-the art in cover song retrieval research, several existing approaches to cover song retrieval are studied in a modular way. Modules from different approaches are systematically recombined and each resulting combination is tested on several designated datasets. The modules have been chosen such that they reflect general and independent system decisions, while the datasets were constructed to each pose a specific and known subset of the broad range of cover song types and similarity challenges. For the experiments that are carried out, we depart from cover song system combinations that use the conventional approach of representing songs as absolute chroma vectors over time. Additionally, we transform these representations into a statistical meta-representation and also study the influences of using relative first-order time-differential information. In the results obtained, several critical choices in system modules influencing the overall performance can be identified. Our work is presented by first discussing the cover song retrieval problem in a top-down way, starting from the general musical and technical issues posed that form the inspiration for our choice of considered system modules. Subsequently, we continue by discussing the existing approaches that were studied in more detail. After explaining our experimental setup and evaluation methodology, the results of our work are presented. Finally, considering the results of our work, several suggestions for future directions in cover song retrieval research are made that especially focus at gaining more understanding in the relation between high-level aspects and technical solutions.

Boosting the performance of a conventional speech enhancement system by applying post-processing restoration modules. The speech production process is modeled with linear prediction analysis (LPA). This yields to a two step problem: enhancement of the spectral envelope obtained after conducting LPA to the output signal of a conventional speech enhancement method and enhancement of the excitation signal obtained after conducting LPA to the output signal of a conventional speech enhancement method.

Motivation: Synthetic biology aims at building biological systems for useful purposes. Relatively simple gene networks have been engineered, but the design process is limited. Many papers advocate the use of engineering concepts like standardization and modular design to simplify the design process and enable the design of more complex systems. Currently, there are no tools available that implement both concepts in a practical way. Results: We have developed a software tool to show how standardization and modular design can be used for the design of logic gene networks. We introduce gene network templates to be able to use modular design in a practical way and use a standard model to simplify the design process and enable reuse of parameters. We have designed three logic gate templates and used them to build two logic gene networks: a demultiplexer and a D-latch. Our software tool was used to turn the templates into devices and to evaluate the performance of the devices. The results show that the devices are evaluated correctly. Furthermore, the results show that for the design of a gene network our method can be used to indicate which biological parts are preferred at what location in the network.

Human machine interaction is not as natural as interaction among humans. That is the reason why is not yet possible to completely substitute face-to-face communication by human-machine interaction in spite the theoretical feasibility of such a substitution in several professional areas including education and certain medical branches. To approach the naturalness of face-to-face interaction, machines should be able to emulate the way humans communicate with each other. Multi-media and man-machine communication systems could promote more efficient performance if machine understanding of facial expressions is improved. Based on this problematic issue, the current thesis is focused on the recognition of emotional facial expressions, finding interrelation between the facial expressions and labels and finally the classification of the expressions. One of the parts of this project is an emotional database which will contain images of faces, their corresponding Action Units and their labels. The contribution of this database to the problem stated above is that it can be used by systems in order to recognize emotional facial expressions given one of the database data i.e. action units’ combination. The other part of the project, which is an expert system for emotional classification, will enable to classify emotional expressions, the ones included in the database and all the possible combination of them.

An increasing number of cameras is being used to monitor a growing range of environments. Consequently, surveillance systems consist of an increasing number of screens to display all incoming video streams, making it difficult for observers to maintain a mental model of spatial relations between videos. A number of systems is developed aiming at improving the observer’s spatial awareness by integrating videos with their spatial context, a 3D model of the monitored environment. In these systems, video content can be viewed from virtual cameras that correspond with the cameras in the real world. To switch between views on certain videos, the observer has to make a transition between the corresponding virtual cameras by some means of navigation. While studying state-of-the-art 3D surveillance systems, we have observed that not much attention has been paid on exploring more sophisticated viewpoint transitions. In this thesis, different classes of camera relations are identified. For each class, a viewpoint transition mechanism is developed with the focus on reducing distortion of the video information during transitions. In order to navigate through the virtual environment, viewpoint transitions have to be initiated to switch between videos. As part of this thesis, a number of concepts have been developed that provide controls to the viewpoint transitions in the form of 3D elements that are added to the virtual environment. We have implemented our navigation concepts in a prototype, which was used for a user study. This thesis concludes with the results from this user study and our vision on future work in the field of context-aware surveillance.

This thesis discusses a new and unexplored concept which uses computer assisted technology to support trauma-focused psychotherapy, focusing on restructuring and relearning of past events. The proposed application allows patients and therapist to visualize the patient’s past experience using maps, personal photos, stories and self created 3D virtual worlds. The tool aims to allow patients to restructure, reappraise and relearn about their past experience involving the problematic stressors. The design of the system followed a situated cognitive engineering approach. The first step of this approach was to do a domain analysis. This was done in close cooperation with a psychiatrist experienced in treating veterans suffering from a combat-related PTSD, which eventually lead to the establishment of an inventory of human factor knowledge, operational demands and envisioned technology. The knowledge was used to create several scenarios and prototypes. Experts with a psychology background were asked to review these scenarios and discuss various possibilities and limitations, while prototypes were evaluated and tested by experts with a background in Human-Computer Interaction. The acquired feedback made it possible to constantly refine the requirements baseline. The experiment which followed suggested that all three main interface components were easy to use. Also, differences were found in a way a story was told with the application compared to a story told without the use of the system. The results hinted at a more structured and precise way of storytelling. A case study with a veteran showed that the patient enjoyed working with the application. He felt encouraged to work with it as he saw the purpose of talking about past events by managing a media archive.

In this thesis the agent programming language GOAL is extended with communication. For GOAL we pur- sue an implementation that has a well-founded theory as well as providing pragmatic programming constructs for the programmer. Existing technologies are reviewed to explore their approaches. Many of these technolo- gies have Speech-act theory as their theoretic base, or build upon other technologies that do so. This led to communication frameworks having vast and unclear performative sets and lacking formal semantics. This thesis goes back to the core of communication by regarding communication from a linguistic perspec- tive. This approach inspires syntactical representation of communication constructs and also leads the way to specifying a formal semantics for those communication constructs. This semantics does not have a receiver of a message refer directly to the mental state of the sender, but rather specifies how a model of that mental state can be deduced from the communication. These mental models are implemented as language constructs which the agent programmer can use to have the agents reason about the beliefs and goals of other agents. Finally the necessary middleware elements are implemented into the GOAL interpreter to allow a system of multiple agents to be distributed across multiple platforms or hosts.

With the increasing demand for ever more depth and detail of modern video games, developers are faced with the problem of how to create and manage large amounts of content. One aspect of this is how to cheaply enable game entities to travel through their virtual worlds in a natural and realistic fashion. Game map sizes and complexity have however risen to such an extent that there is a strong demand for automization in order to relieve artists and designers from their medial tasks, and enable them to focus more on the creative aspect of game design. As a solution to this, we introduce systems for the autonomous generation of precomputed Navigation Meshes (NavMeshes) and their in-game application. These meshes contain abstractions of all walkable surfaces of a static map environment in the form of a set of convex areas and a matching graph topology. We discuss the pros and cons of generating them solely from map collision volumes using a 'lightweight' form of Boundary Representation (B-rep or BREP) algorithm, to help remove areas that cannot be reached due to the dimension of the traveling objects. This B-rep approach provides compact yet accurate NavMeshes abstractions that are ideal for classic path-finding algorithms. Natural movement around dynamic obstacles is achieved using a combination of 'fuzzy' whiskers sensory systems, and a 'deterministic' fall-back mechanism that temporarily enhances the resolution of the NavMesh graph locally. Further speed-ups are obtained by parallelizing classic A* algorithms for nowadays common multi-core architectures. We introduce the 'Parallel Bidirectional Search' that significantly outperforms traditional A* implementations.

Assume that two sub-graphs of a graph are available and that the graph represents the corridors in a building. Without using any previous knowledge on the entire graph, can these two sub-graphs be used to construct an extended view on the entire graph. To test this assumption a collection of algorithms will be designed and implemented which can fuse two sub-graphs (topological maps) together to achieve this extended view on the unknown environment. The main goal for this project will consist of the distributed construction of topological maps within a simulated environment, with the main focus on the used algorithm(s) for the fusion of multiple maps. This will be realized with the help of a “graph” data structure, a multi agent system and match and merge algorithms. In general the simulated environment will function as a test environment for the algorithms. A real life situation for the simulation environment could be, that it represents a building in which an emergent situation has occurred and in which individuals are exploring the unknown layout of the building in search for their personal goal(s). This system can be used by first responders who are exploring the building and are in search of casualties in case of a fire or a terrorist attack. This will all result in a simulation in which it is assumed that each explorer in the field is equipped with a Personal Digital Assistant (PDA) or a different handheld device and can communicate with the other explorers within its communication range, without the help of a centralized communication system. The explorer’s traveled path is stored into their personal PDA and the software running on the PDA’s shows the explorers their traveled path and if possible combines their map with the received map from the explorers they encounter. If a combination is possible the software on the explores PDA executes the algorithms to construct the most complete view on the environment as possible. The completed or partially completed map can also provide guidance to the explorers if this is needed.

With the advent of affordable and powerful game engines, simulation games have become an increasingly effective solution for organising virtual training sessions in areas such as incident response, safety supervision and medical services. These simulation games are replacing real-life training sessions, which are often expensive and time-consuming. The downside with using simulation games for training purposes is that current simulation games lack the flexibility of adjustable training programs that real-life trainings have, due to simulation games being a finalised, and thus static, product. This research focuses on how to increase the flexibility of training programs offered by simulation games. This is done by abstracting the scenario of simulation games, so it can be re-arranged by the training instructor using a standalone editing application. A prototype of such an application, called Shai, is developed to demonstrate and validate this approach.

Beat³ is een computerspel voor alle leeftijden, waarbij de speler woorden moet vinden in een woordveld en deze zo snel mogelijk moet selecteren. Omdat dit spel voor de Game Development competitie van Microsoft's Imagine Cup is gemaakt, zijn de acht millenniumdoelen verwerkt in het spel en heeft deze een sterk educatieve waarde. Daarnaast is een sterke gameplay, gebaseerd op visuele en vooral auditieve terugkoppeling één van de sleutelelementen van Beat³. Beat³ is gemaakt met Microsoft Visual C# 2008, Microsoft XNA Game Studio 3.0 en de Cannibal Game Engine. Beat³ is zowel speelbaar met het toetsenbord als de Xbox360 controller en kan gespeeld worden door meerdere spelers. Tijdens het ontwikkelen, hebben we gebruik gemaakt van de Scrum methode. Deze is vanwege ons complexe studieschema niet geheel tot zijn recht gekomen. De deelname aan de Imagine Cup is gestrand in de halve finale, maar heeft wel een mooie leerzame ervaring voor het team opgeleverd.

A new Group Decision Support System (GDSS) has been developed that simplifies and improves group decision-making for travel consumers. Prior studies have shown that various irrational factors influence a group decision-making process. Moreover there is a lack of structured information that supports the decision process. Presently, different studies have proposed group recommendations systems that improve the decision-making process. However, they appear ineffective in leading a group towards a satisfying outcome. The objective of this research is to develop and evaluate an intelligent group recommendation mechanism that is integrated in a well-structured decision process. As a result, a prototype of an online intelligent group decision support system for travel consumers has been built, which we refer to as Trip.Easy. Subsequently, we set up a user experiment and invited over a 120 participants, divided into 30 groups. Each group was instructed to organize a city trip while using Trip.Easy. After each session, the participants were asked to fill out a questionnaire. The results show that users were satisfied with the outcome of the session and with the recommendations of Trip.Easy during each session. Furthermore, analyses show that even groups with a relatively high measure of conflicting preferences, are guided towards a satisfying outcome. Finally the results show that the satisfaction with the recommendations significantly increases as the session continues. We conclude that Trip.Easy is able to lead a group towards a satisfying outcome.