18-12-2010, 03:24 PM
Project Oxygen
Introduction
For the past six months, I have been integrating several experimental, cutting-edge technologies developed by my colleagues at MIT as part of the MIT LCS/AIL Oxygen project. This paper gives a snapshot of this work-in-progress. Project Oxygen is a collaborative effort involving many research activities throughout the Laboratory for Computer Science (LCS) and the Artificial Intellegence Laboratory (AIL) at the Massachusetts Institute of Technology (MIT). The Oxygen vision is to bring an abundance of computation and communication within easy reach of humans through natural perceptual interfaces of speech and vision so computation blends into peoples’ lives enabling them to easily do tasks they want to do – collaborate, access knowledge, automate routine tasks and their environment. In other words, pervasive, human-centric computing. At first blush, this catch-phrase appears vacuous. Today, computers are certainly pervasive; it is likely, at this moment, you are within 100 meters of a computer. Computers are certainly human-centric; what else can they be? On the other hand, computers are not yet as pervasive as is electricity or water. Although computers perform jobs required by humans, they do not feel human-centric – humans must conform to an unnatural way of communicating and interacting with computers. Finally, the tasks described have little to do with computation; computermediated functions is a more accurate term but sounds awkward. The vision and goals of the Oxygen project are described in detail elsewhere [11, 2, 1], the purpose here is to show how many maturing technologies can be integrated as a first step towards achieving the Oxygen vision. There are research efforts at other universities and research institutions that roughly share the same vision, however, each institution focuses on integrating their own maturing technologies. Oxygen has a three-pronged approach by dividing the space into three broad categories: the H21, a hand-held device, the N21, an advanced network, and the E21, a sensor-rich environment (see Figure 1). In what follows, an Oxygen application is described in terms of its human-centric features as well as the required technologies. It is important to keep in mind that this is just one of many applications and that it is merely a vehicle to explain how many technologies can be integrated and how to create the infrastructure necessary to enable the introduction of context into applications making them more “natural” to use. The sample application is that of a seminar presentation support system. The next section gives an overview of the application. Section 3, reviews many of the technologies that will go into this application. Section 4, shows how they integrate to form the application. A preliminary programming language and middleware support is described .
Acomputer-mediated, seminar presentation system
This section describes a computer-mediated seminar presentation system. As you read through the description, compare it to how presentations are given today. Although a laptop with programs like Powerpoint or Freelance attached to an LCD projector is a vast improvement over the old days of foils or 35mm slides, the human has given up a degree of control, freedom, and naturalness. The system described below provides for a more natural human interface. Alice is to give a seminar about her O2.5 project. As she walks into the seminar room, she allows herself to be be identified as the speaker. She does not need to carry a laptop with her slides on it – all of her files are globally accessible. Alice tells the system how to find her talk by simply supplying enough keywords to uniquely identify the file she wants. Her files are well indexed and so she merely describes the file in human terms and not with some bizarre syntax. The system knows where she is and marshals all the physical components that may be needed for her to control the display. Alice wants to control the display so that it matches her current desires. A seminar is a live event and the dynamics depend on the audience and speaker. Although it is crucial that she control the presentation, this control should be of minimal distraction. The same is true for the audience – they should be able to see the visual content, hear her commentary, and take notes at the same time. Moreover, unexpected events should be handled in a natural way. Even today, Oxygen technologies can make a computer-mediated presentation a more natural experience. In particular, three natural ways to control the slides are provided, as opposed to the traditionally way where Alice either clicks the mouse or hits the enter key. It is computer-centric to force the speaker to always walk over to the laptop in order to advance the slide. A wireless mouse is only a partial solution as it requires that something be held in a hand. For Bob this might be fine, but Alice likes to use a laser pointer to highlight objects on the screen and she finds holding two objects to be very awkward. An integrated pointer/mouse is no better since it now requires attention to find the correct button. Alice can use her laser pointer to highlight words, draw lines and sketches, as well as to switch between slides. Holding the pointer in the bottom right corner means to advance to the next slide. A camera looking at the screen interprets Alice’s laser pointer gestures. But not all humans like to use laser pointers. Some people, especially when they
are continuously engaged in speaking, like to use verbal commands to control the presentation. This is done with a microphone and software that continuously tries to understand commands. All three modes of control will always be active, allowing the speaker to use whatever is convenient.
There is more to a presentation than just advancing slides. Alice may want to see her notes, see the next slide before the audience, skip a slide or present the slides in a different order. A laptop, handheld, or any other personal communication device can be used by the speaker. To skip to a different slide without anyone knowing it, is a task that is easily performed by simply clicking on a different slide image on her personal display. The personal display must remain consistent with the public display. So, whether Alice says “Next Slide,” chooses a slide from her private computer (handheld or laptop), or uses the laser pointer, both displays are updated.
The audience should also have a choice of ways to observe the presentation. They can look at the large projection screen in the front of the room, as is usually the case, or they may choose to view the presentation on their own personal digital device. The output is simultaneously broadcast to these devices. Some people in the audience might like to take notes and have them correlate with the presentation itself. We propose broadcasting a URL or some other identification symbol for the current contents. This can be either used to display the slide on the laptop, or be inserted into their notes. Later on in the privacy of their own room, these notes can be merged with an archived version of the talk. The archived version will match the presentation rather than the original file. Alice may have many “emergency” slides prepared that will be shown only in response to a question. To summarize, there are several output modalities: the LCD projection, a broadcast of the current content, an archival copy that can be accessed afterwards, and the ability to correlate the public presentation with her own personal view of the presentation.
Lastly, Alice also has “meta” operation control - e.g. switching to a different presentation package, such as a browser or Mathematica, or even to the contents of another presentation. She should also be able to control whether or not content is broadcast or archived.
Technology overview
Research into many technologies that support the above scenario being pursued as part of Project Oxygen. Once again, we wish to emphasize that there are many competing technologies being developed elsewhere. We deliberately ignore them for several reasons1. First, to do justice to them all would make this article too large. Second, close physical proximity is usually required when making use of experimental, research systems. While it is possible to do this remotely for one component, it is nearly impossible do this for a number of research projects. We wish to provide feedback to these other research projects before they are ready for prime-time and we deliberately try to use them in some unintended way. While there are similar efforts in many of the intelligent or instrumented rooms, our example is simply geared towards exposing how components interact even with commodity hardware. As fun as it is, the particular demo of an oxygenated presentation is not the goal
For more information about this article,please follow the link:
http://ai.mit.edu/projects/oxygen/oxygen...oll-02.pdf
