Monday, September 26, 2011

Intel Labs' 5th Science and Technology Center will focus on Pervasive Computing

Intel Labs has announced that they will establish a fifth Intel Science and Technology Center (ISTC), to be hosted at the University of Washington, which will focus on research in Pervasive Computing.The ISTCs are part of a $100M program to invest in U.S. university research, which was announced by Intel in January 2011. Previously announced ISTCs include the center for Visual Computing at Stanford, Secure Computing centered at UC Berkeley, an ISTC for Cloud Computing with a hub at Carnegie Mellon University, and  the ISTC for Embedded Computing, also at Carnegie Mellon University.

Intel commits funding to the ISTCs for a period of five years (with a checkpoint at three years), along with assigning four Intel researchers per center that will work directly on the hub campuses. The ISTC collaborative model includes a set of spoke universities along with each center hub. The plan is for any intellectual property that is developed to be put into the public domain, and for any software that is developed to be open source. The co-principal investigators for the Pervasive Computing center are Dieter Fox, associate professor of Computer Science & Engineering at the University of Washington, and Anthony LaMarca, a senior scientist at Intel.

The Pervasive Computing team has chosen three research themes that will be the focus of their work:
  • Low-power sensing and communication
  • Understanding human state and activities
  • Personalization and adaption

To enable low power, unobtrusive sensing and communications, the ISTC plans to employ energy harvesting techniques, and to develop energy-efficient network protocols.

The goal of the second pervasive computing theme, understanding human activities, is to be able to "recognize a user’s fine-grained context and interactions with people, objects, and environments", said research Dieter Fox. Fox envisions the integration of a heterogeneous set of sensors with extensive use of cameras, integrated into mobile devices, and embedded in a user's home and the environment. During an analyst briefing preceding the announcement, Fox cited the Microsoft Kinect sensor and camera system as an example of a device that can be leveraged to extract more complex human context, through its ability to sense depth. In order to make pervasive computing real-time and efficient, the researchers will investigate how to partition execution of algorithms between mobile devices, embedded computers, nearby servers, and cloud computing.

For personalization, the goal is to have pervasive computing systems that will be interactive, and continuously  learn user’s preferences, their environments, and routines, etc. The notion is for personalized models in the system to have the capacity to be taught new activities by example and demonstration. On this last theme, the researchers will need to balance on a fine line between pervasive computing and invasive computing. In a whitepaper that was published as part of the announcement, the authors note that ensuring security and privacy is a key issue in their work.

Exploring new technologies to support the next generation of pervasive computing will require dealing with significant quantities of private information. To ensure the trustworthiness and security of the systems involved and to safeguard privacy, security and privacy researchers will be involved in all of the center’s research efforts. They will serve as internal consultants who will proactively surface, assess, and address potential privacy concerns related to the technologies being explored. They will also strive to develop a deeper understanding of what trust and privacy means for each problem domain, and to develop new technological solutions for achieving trustworthiness and privacy when standard best practices are not sufficient.(From whitepaper on The Intel Science and Technology Center for Pervasive Computing).
The Pervasive Computing smart cooking assistant would have the ability to learn recipes by
observation with multiple depth-sensing cameras.

Intel's Anthony LaMarca said that the Pervasive Computing ISTC team is developing concept applications to drive their work in three different domains: mobile health and well-being, the smart home and family coordination, and task spaces.

In an example of a smart cooking assistant, LaMarca described a kitchen-oriented task space that would have knowledge of recipes, be able to keep track of cooking progress to provide reminders and guidance, interact with cooks via audio, gesture, projected imagery, and use multiple depth cameras to learn detailed nuances of new recipes by demonstration.

The example application of family coordination may evoke feelings of Big Brother, with its goal being to "help families coordinate their busy lives through monitoring, tracking and reflecting on their activities". LaMarca said that by following a family through their daily routines, such a pervasive computing system would advance the state of the art in machine learning to unsolved problems such as task interruption and task coordination, for a whole home or organization.

The mobile component of the pervasive computing project is focused on personal stress, with the goal to "help users identify, manage, and reduce stress and anxiety in their daily lives". The hope is to identify the causes of stress by using sensors in a mobile device to follow a user's daily routine. LaMarca said that crowd sourcing of data would also be leveraged to help users manage stress.

In response to the inevitable concerns for machines replacing human interaction, LaMarca said that capturing expert domain knowledge, such as for cooking, would enable bringing it to the masses in a more detailed way. Perhaps we all could have Julia Childs or Rachel Ray as kitchen companions to guide us in preparing gourmet meals. LaMarca also said that he believes that the real time learning, adaptation, and personalization capabilities of pervasive computing can be an enormous boost to personal health and wellness.

No comments: