 |
|||||||||||||||||||||||||||||||
|
|
e-World Lab Honours and Software Engineering Projects for 2003
This is a proposed list of projects in the e-World Lab available to qualified Software Engineering and CIS Honours
students in 2003.
This project offers an opportunity to work with a team of DSTO software engineers on the InVision component-based software framework for information visualisation. The framework is implemented in Java, and exploits technologies such as Swing, Java2D/3D, XML, ODBC and OQL. Experience in the use of open source software project management technologies incorporated in the SourceForge tool suite is also anticipated. The project involves the design and development of user-interface components for a flexible enterprise modeling tool for the InVision component infrastructure. User interfaces for the visual design of the meta-model (schema) and for entry of the model data are required. Research would need to be carried out into an effective interface for the data entry component. It is anticipated that the Meta model designer would use a similar paradigm as existing UML class diagram editors such as Rational Rose (perhaps drawing from the work being done on enterprise modeling languages by the Collaborative Research Centre for Enterprise Distributed Systems Technology – DSTC). The work would promote important interaction with and contribution to research being undertaken at DSTO and the newly created DSTC SA node.
Project Type: Software Engineering Honours Ubiquitous computing relates to a future where computers will become woven into the backdrop of our everyday environments. The LiveSpaces Project within e-World Lab is experimenting with the development of future enterprise-enabled ubiquitous workspaces. These future working environments will be context aware, use multiple displays and interaction devices, and host a range workspace services and applications that will support natural interactions such as the use of gestures, speech, and multimodal interfaces. Computing applications will move from being hosted on workstations to being hosted within physical spaces such asmeeting rooms. InVision is a novel model-based visual information system developed by DSTO and other partnering organizations. It supports the presentation of a range of visual representation across various devices including large-screen displays, tablet PCs, workstations, and augmented reality. This project will involve hosting InVision within the e-World Lab LiveSpace. Participants will learn about next generation operating environments such as Stanford’s iROS, Java2D/3D, augmented reality, ubiquitous computing devices, and visualization technologies. Research will need to be conducted into the architectural aspects of ubiquitous computing applications. The work will involve significant interactions with e-World Lab and DSTO researchers.
Project Type: Software Engineering Honours LiveSpaces is a project within the e-World Lab that is undertaking research into future workspace environments. These future workspaces will use ubiquitous computing infrastructure to support natural interaction between people, technologies and media. DSTO has developed an approach called AuTM for the real time transcription of meetings, interviews or brainstorming sessions. During a meeting AuTM transcribes spoken utterances into text, logs start and stop times and saves the audio. AuTM also allows annotation such as adding highlights, motion and action items. The prototype currently requires one computer and microphone per speaker. AuTM will provide the basis of a new Automated Transcription Service for LiveSpaces. This project will require the continued development of AuTM to allow multiple speakers to use only one computer. This will involve four subtasks:
Participants will gain significant expertise in the areas of speech recognition, ubiquitous computing, and software development approaches. The work will involve substantial interactions with DSTO and e-World lab research staff.
Project Type: Software Engineering Honours LiveSpaces is a project within the e-World Lab that is undertaking research into future workspace environments. These future workspaces will use ubiquitous computing infrastructure to support natural interaction between people, technologies and media. DSTO has developed an approach called AuTM for the real time transcription of meetings, interviews or brainstorming sessions. During a meeting AuTM transcribes spoken utterances into text, logs start and stop times and saves the audio. AuTM also allows annotation such as adding highlights, motion and action items. The prototype currently requires one computer and microphone per speaker. AuTM will provide the basis of a new Automated Transcription Service for LiveSpaces. Elvin is a messaging/notification system developed by DSTC that allows a user to access a subscription service. Elvin messages are routed to one or more locations, based entirely on the content of each message, and can be used for communicating with components over the Internet. Elvin is a key component of the LiveSpaces infrastructure. AuTM currently uses its own messaging protocol. Elvin appears to provide several advantages in the areas of computer connectivity, distribution, system resource utilization and message security. An Elvin Client will need to be developed which interfaces with the AuTM messaging component. Participants will gain significant knowledge of speech recognition, the conceptual basis for speech transcription services, publish subscribe infrastructures, and ubiquitous computing. The work will involve significant interactions with DSTO and e-World lab research staff. Project Type: CIS Honours LiveSpaces is a project within the e-World Lab that is undertaking research into future workspace environments. These future workspaces will use ubiquitous computing infrastructure to support natural interaction between people, technologies and media. In this Honours project, the student will continue, within the context of the e-world lab LiveSpaces, the design and development of a spoken dialogue system which was started over the summer 2002-2203 in the context of a spoken dialogue system for a Virtual Agent in the DSTO Future Command Centre Analysis Laboratory (FOCAL). The project will take as starting point the dialogue system which was set up and deployed in the FOCAL environment during the summer project, and will deploy it as a LiveSpaces application in e-world lab. The project itself will consist of the design and development of a new dialogue system, specific to the LiveSpaces context, e.g. requiring the environment to perform a certain task or set of tasks, or requesting information to be given verbally or displayed. This involves acquiring an understanding of dialogue design, and of the integration of speech recognition and TTS components. The project will also involve the design and development of reusable grammars. These grammars must be tailored for specific goals, such as listening for particular requests (e.g. to display some information, or to perform a certain task), or giving specific information (e.g. specific times or locations). As the Natural Language Understanding (NLU) component uses the Regulus Opensource software, this require some knowledge of, or willingness to learn to use, Prolog. Although it would be helpful if the candidate already had an understanding of the limitations of speech recognition and of the integration of speech within dialogue systems, this is not required. However, the successful candidate can expect to have obtained such an understanding by the end of the project. Being part of a team working on a larger project will also provide the student with excellent opportunities for understanding interconnections with other fields (speech, AI visualisation, reasoning) as well as the opportunity to develop teamwork skills
Project Type: CIS Honours The LiveSpaces Project within e-World Lab is experimenting with the augmentation of physical spaces (such as meeting rooms) with ubiquitous computing technologies to provide the basis for the rapid development of future Interactive Intelligent Workspaces. These future workspaces will use ubiquitous computing infrastructure to support natural interaction between people, technologies and media. In these environments computing applications will move from being hosted on workstations to being hosted within physical spaces such as meeting rooms. This work extends research undertaken at Stanford University in the development of operating environments for ubiquitous computing rooms. For example, Stanford researchers have developed the concept of a room controller. The proposed project is based on work undertaken by e-World Lab intern students during the summer 02/03 to extend the concept of a room controller into a full desktop-type metaphor for an interactive intelligent workspace. For example, the room desktop would allow users to control various devices such as lights, projectors through a graphical representation of the room as is done by the i-Crafer Room Controller. It would also allow drag and drop capabilities to, for example, associate particular files/information/media with one of the many room devices (eg large screen display, Tablet PC, etc) and have it presented by way of an appropriate application. For example, a user might drag a Powerpoint file from the room desktop (displayed on their table PC) to a smartboard display for presentation. Other participants might use the Room Desktop to have the same file be presented on their own Tablet PC so that they can view the presentations notes or look ahead. Other aspects to be explored include the concept of a room clipboard.
Project Type: Software Engineering Honours The LiveSpaces Project within e-World Lab is experimenting with the augmentation of physical spaces (such as meeting rooms) with ubiquitous computing technologies to provide the basis for the rapid development of future Interactive Intelligent Workspaces. These future working environments will be context aware, use multiple displays and interaction devices, and host a range workspace services and applications that will support natural interactions such as the use of gestures, speech, and multimodal interfaces. Kayla is an embodied virtual assistant who supports activities being undertaken within a LiveSpace. She will provide announcements, answer questions, retrieve information, and control the workspace environment. Initial Kayla concepts were explored as part of a summer intern project. This follow-on project will undertake underlying research into embodied virtual assistants, develop Kayla concepts and develop a Kayla prototype implementation based on an existing development kit such as that provided by 3D-SeeMeNow technology. This prototype will be integrated into the e-World Lab LiveSpaces environment and evaluated.
Project Type: CIS Honours InVision is a model-based visualization system being developed by DSTO and other partnering organisations. InVision uses the concept of an underlying composite model from which various visualizations can be generated. The model comprises a Meta data description and a graph-based instantiation which captures details of sets of target entities and the relationships between entities. Models can be generated from databases, hand developed in xml etc. The modeling approach is generic and can be applied to the modeling and visualization of many domains such as software visualization, the visualization of transport systems, social networks etc. Currently InVision allows for the management and access of models but does not allow for fusion of models. This project would look at approaches for fusing InVision models. This would involve the fusion of metadata/ontology descriptions and methods for addressing contentions. The student would be required to undertake research into appropriate fusion methods and then implement a component for use in InVision. The project would involve substantial interaction with DSTO personnel and e-World lab researchers.
Project Type: CIS Honours Some of the characteristics of ubiquitous, future workspace settings, such as LiveSpaces, are new environmental interfaces, increased use of large displays and the reduced reliance on keyboards and mice. Users will interact more naturally with technology and these future workspaces will require lower level support in the form of generic workflows and process automation. In addition, Ubiquitous computing environments are often made up of many disparate systems that support different tasks and services and these need to be integrated and coordinated to achieve goals at many levels within an enterprise. This project concerns itself with creating an interface between the user and ODSI (Open Distributed Systems Infrastructure from DSTC), which is a peer-to-peer network. It will involve building a generic API in Java to interact with ODSI, a middleware layer that is written in Python and provides enterprise services such as workflow generation. The user interface to this API will allow a person to rapidly specify, create and manage workflows to achieve enterprise goals. These interfaces could link to multi-agent systems, other workflows, voice-activated commands, virtual advisors and a number of other systems to create technologically rich, yet natural user environments. The student in this project will gain valuable exposure to ubiquitous computing concepts including programming in a peer-to-peer environment, the application and use of middleware, workflow management techniques and an understanding of the enterprise model. The student will also acquire useful experience with Elvin, a content-based message routing service, which forms a key underlying component in the LiveSpaces architecture.
Project Type: CIS Honours Wireless infrastructure networks provide access to the wired network via a series of access points. Ad-hoc networks involve the mobile nodes directly communicating with one another to form wireless network. Multi-hop ad-hoc networks are a hybrid approach where each of the wireless nodes as a router to extend the coverage of the infrastructure based approach. The goal is to increase capacity, coverage while reducing overall power consumption. The aim of this project is to investigate algorithms for the fragmentation and replication of files across such hybrid environments. Aspects of this project will include I) a survey of existing components of multi-hop wireless ad hoc networks, such as routing (eg. DSDV, DSR), fragmentation z(at Transport and Application layers) II) the development of a simulation environment for such large scale hybrid networks III) the development of an application layer model for information fragmentation and replication IV) the development of a series of methods and measures based on this model V) an performance evaluation of this model under varying connectivity conditions. This project involves a collaboration with the pervasive computing group at sydney university.
Project Type: We have extended InVision with components that allow any of the views in an InVision workspace to be displayed through an AR tangible interface. The AR view capability was added by incorporating the ARToolkit Version 2.52 (Kato, H. and Billinghurst, M., 1999) as a component into the InVision framework. A view is selected from an InVision workspace for display as an active view on the workstation screen. Right clicking on the active view tag allows the user to also display this view in AR mode relative to a particular fiducial marker pattern. The user can then inspect the model by rotating it, changing the azimuth etc. This is particularly useful for three-dimensional models. By rotating the marker, the user can examine the model from all viewpoints. By bringing the marker closer or further from the user, their viewpoint can be zoomed in or out. This provides an interaction that can be closely mapped to the natural interactions between humans and the objects that they want to analyze. The following is a list of goals that MARVIS was designed to address.
A fuller description is available on request . please email Thomas@cs.unisa.edu.au
Project Type: The Tinmith outdoor augmented reality system (developed at the School of Computer and Information Science Wearable Computer Lab) is currently used for complex outdoor modelling tasks, allowing users to add 3D geometry to the real world and make changes interactively. Using this application, users can preview what objects will look like before being physically constructed, and there are a number of potential applications for surveying, construction, and design areas which we are beginning to explore now. To interact with the modelling system, the user wears a mobile backpack computer, a head mounted display, and uses a user interface based around a set of pinch gloves, with tracking provided by a video camera mounted on the head. Currently, these gloves are tracked using a library known as ARToolKit, and is used to provide 2D cursor tracking for working with the 3D environment. This software toolkit is CPU intensive, and for small machines such as handheld computers, is not possible to run in real time. As a result, we would like to implement a vision tracking system in hardware, thus reducing the CPU usage. As there is no .off the shelf. application specific integrated circuit (ASIC) to complete such a tracking problem, a dedicated hardware implementation has to be developed. Developing a custom ASIC can be a very expensive exercise, in both time and money. With Field Programmable Gate Arrays (FPGA) it is possible to prototype such a hardware implementation without having to spend money on expensive mask costs as the design can be progressively improved and tested. This is possible because FPGAs are configured in software, but still gain the hardware speedup of an ASIC. Designing hardware circuits has traditionally been left to only experienced hardware designers. With the development of new hardware description languages such as HandelC (a C like hardware programming language), it is possible for software engineers to design and implement hardware with little background knowledge. It has also made the problem of converting software implemented algorithms into hardware much easier. Therefore we believe this project is very suitable for a summer intern student. Rather than performing full six degree of freedom tracking like ARToolKit, only simple 2D tracking is required, simplifying the initial task. Initially, the camera on the head (currently used for the AR overlay) will be used to supply video data to a laptop PCMCIA FPGA card. By attaching uniquely coloured balls to the gloves, the hardware algorithm executing on the FPGA will detect these markers and return their coordinates to the laptop for further processing. In future, we believe it should be possible to take the video stream live into the FPGA without the CPU's involvement, meaning live video tracking could be done on even the most primitive computers with no overheads. A tracking system based on FPGAs could potentially be manufactured very cheaply, and emit a serial data stream which could be interfaced with any number of devices and used for tasks not originally indented by the designers. Also, another future project would be to convert ARToolKit itself into an FPGA design, and this project would be a proof of concept and to gain the experience needed for this much more complicated task. As part of the project the student would be expected to write the hardware tracking application, under the supervision of staff from both the Wearable Computing Lab and the Reconfigurable Computing Lab. This would include loading the video data stream onto the FPGA platform, writing the hardware implementation of the tracking algorithm (currently written in software) and returning a set of coordinates via a serial connection. Once the project is complete, we foresee at least two international conference papers (one in the FPGA area, and one in the AR area) based on this work, as it is the first time this work has been performed in either environment, and so this system would be pioneering work in the field.
Project Type: This project is geared to improving the administrative effectiveness of workgroups, using the School of Computer and Information Science (CIS) as a test bed. It involves a network of cooperating software agents, including a workflow service. Nearest to the user is an agent that helps each individual learn about the processes, acts as a wizard to assist the user in starting new cases of these processes - or in creating new processes, and provides help on request. It depends not only on access to the library of processes (see below), but also on user profiles indicating what are each user's special needs, priorities and default methods of working. The intelligent workflow agent will act as a prompter to all staff in the school, providing them with a task list of the things they need to do at various stages of the academic year or on special projects. Hopefully, it would relieve the HOS and delegated senior staff from having to personally field all the administrative workload. One subtask is to capture and encode all the process rules, intelligence, logic, data etc that represents the School's administrative knowledge. Another task is to create a "Help" facility in hypertext form to provide induction material to new staff or staff new to a role. This project would not involve building a new workflow management system from scratch, but we envisage enhancements to the FlowMake/Chameleon system from DSTC that we are currently evaluating.
Project Type: |
||||||||||||||||||||||||||||||
