Pictures of the Future    Fall 2006

Converging Trends and A Holistic Vision of Data

With a total staff of 430, Siemens Corporate Research in Princeton, New Jersey, is Siemens’ largest research institute outside Europe. For more than 25 years, it has been responsible for producing top quality innovations that are incorporated into a full range of Siemens products and services. Today, it is a leading research organization for medical imaging and real time vision.

For Paul Camuti, head of Siemens Corporate Research (SCR), the focus of his eight departments boils down to one thing: a new vision of data. Data, as he sees it, is embedded in what sensors see—an approaching vehicle, an unclaimed briefcase in a rail station, a hearing aid on a production line, or the indistinct outlines of a beating fetal heart in an ultrasound scan. "The vision," says Camuti, "is that although we have a range of sensing technologies that deliver images for purposes ranging from automotive safety and security to automation and medical diagnostics, in the end, all of the images are just pictures of data, and once we grasp that, we will begin to manage that data synergistically and extract a new level of information from it."

At SCR, with its 280 scientists developing technologies in the areas of automation & control, imaging & visualization, intelligent vision & reasoning, integrated data systems, real-time vision & modeling, software engineering, user experience, and knowledge management, that vision of a grand convergence of data and an ensuing world of concise, actionable information is still just a gleam in Camuti’s eyes, yet it draws closer each day as computing power grows.

Examples of converging data flows are coming up fast. In the automotive safety area, for instance, SCR researchers are working with Siemens’ SV automotive Group and its customers to develop a range of technologies such as traffic sign recognition, advanced cruise control, night vision pedestrian detection, and driver monitoring. Known as "pro.pilot" (see Pictures of the Future, Fall 2005,  Driver Assistance), these vision-based driver assistance technologies will come together in the future as robust data fusion systems are developed. "Eventually," says Camuti, "they could appear as a package, and one day, thanks to such systems, we may have the option of being able to switch our cars to autonomous driving."

Journey Through the Human Body. A similar dynamic is taking place in the healthcare field. For instance, at Johns Hopkins University in Baltimore, Maryland, researchers from Siemens Corporate Technology and Siemens Medical Systems are developing software that will help cardiologists and radiologists not only visualize the location of a catheter in real time as it moves through the body, but also map it into a pre-operative CT image, thus allowing pinpoint accuracy for a variety of treatments ( Johns Hopkins Institute, and Pictures of the Future, Fall 2005,  Trends Digital Health). These steps are opening the door to, for instance, safe elimination of cardiac arrhythmias without medications, pace makers or open heart surgery. "What we’re seeing in such procedures," explains Camuti, "is data convergence—the convergence of what were once the separate worlds of real time and post processing information, the convergence of data from different imaging modalities, the convergence of diagnostic and therapeutic techniques, and the convergence of imaging and non-imaging data in fields like medical informatics. All of this will help to make a range of treatments faster, safer, and more effective while cutting hospital stays, thus reducing costs and improving medical outcomes."

Examples such as these provide insight into another kind of convergence being promoted by Camuti: A convergence of interests between SCR researchers, their immediate customers within the Groups, and end customers, be they medical centers, rail authorities, or manufacturers interested in the most advanced automation and control technologies. All of this represents an expanded focus for SCR. "Our goal, " says Camuti, "is to become not only a leading research organization for medical imaging and real time vision within Siemens, but within the research community at large." This ambitious vision has practical roots. "Although our charter is to develop technologies for the Groups," explains Camuti, "innovation itself is getting to be more and more of a collaborative process. That means being in touch with lead customers, federal research programs, and universities. It boils down to broadening the community we tap for innovative ideas.

" With this in mind, SCR has established more than twenty collaborations in the past year with leading institutions such as Johns Hopkins, MD Anderson, Carnegie Mellon University, Georgia Tech, and Virginia Tech.

And that collaborative community is growing steadily. In addition to the extraordinarily diverse mixture of talents and cultures represented by SCR’s professional staff, the organization benefits from its Berkeley, California-based Technology-to-Business Center, where a team of venture technologists continuously scouts universities, small start-up companies and other innovation sources for new inventions, different approaches, and radical ideas, bringing the best of these to Siemens.

SCR is also benefiting from a convergence of interests on the international level. At CT India in Bangalore, for instance, where SCR has incubated a number of projects, a core competence has emerged in machine vision hardware. "Here in Princeton, our focus on vision has been in algorithms, software and architecture. But we did not have a research agenda designed to look at trends and technologies in related hardware," says Camuti. Bangalore, on the other hand, offers talent, design support, and manufacturers that have specialized in this specific area. By connecting the dots between Princeton and Bangalore, SCR has come up with a more holistic approach to vision system development—one that will eventually cover a more complete set of the Groups’ future needs worldwide.

And what might those needs be? Whether vision systems are used to make our homes responsive to our needs, to guide our cars autonomously, or to make our airports, cities and subway systems safer, they will need to function as a network. For that to happen, each system will have to have the capacity to extract key information from what it sees, and pass only that information on to data collection nodes. "I would say that what this boils down to is the beginnings of a revolution in terms of intelligent video processing—in short, a holistic vision of data," says Camuti.
                                                                                                               Arthur F. Pease