Research

The analysis of ubiquitous biometric data is fundamental to pervasive, human-centric computer technologies. A major technical focus of our research is the extraction of human perceptual signatures for recognition. Our work exploits novel machine learning techniques inspired by mathematical methods developed in fields such as psychometrics and econometrics. Not only is biometric computing relevant to the burgeoning security industry, but when directed at ourselves, biometric technologies can serve as reflectors that enhance our self-awareness, understanding, and health, and they can facilitate our interaction with each another and computers.

We have built a microfluidic logic family and architecture based on two-phase flow. In this scheme, presence or absence of a bubble or a droplet in a fluidic network represents a bit. Logic operations are executed based on direct or indirect bubble interactions. The system is sufficiently nonlinear to exhibit universal logic. The nonlinearity is derived from boundary conditions of two-phase flow. With the emergence of very large-scale integrated microfluidic systems, there is an immediate need to build local flow control systems. Bubble Logic can be used for various applications including bubble display, combinatorial synthesis, and control architecture for lab-on-chip devices.

The CalTech/MIT Voting Technology Project was established by Caltech president David Baltimore and MIT president Charles Vest in December 2000 to prevent a recurrence of the problems that threatened the 2000 U.S. Presidential Election. Specific tasks of the project include evaluating the current state of reliability and uniformity of U.S. voting systems; establishing uniform attributes and quantitative guidelines for performance and reliability of voting systems; and proposing specific uniform guidelines and requirements for reliable voting systems.

For various reasons, some communities have good connectivity amongst themselves, but limited or inadequate contact with a broader community. The CLAP (Cross-Layer Algorithms for Phealth) project is exploring ways to monitor quality-of-life parameters. We are developing low-cost, low-bandwith sensor networks that support applications for capturing, interpreting, and communicating various aspects of both community data (e.g., water, air, or soil quality), and individual data (e.g., heath, hygiene, nutrition), in settings where there is better connectivity among group members than to a core network. Examples of such communities may be found, for example, in OLPC deployment scenarios.

Conventional 3-D printing processes are material-dependent and irreversible. We are working on an alternative approach based on 3-D assembly of mass-produced 2-D components of digital material. This significantly enlarges the available material set, allows reversible disassembly, and imposes constraints that reduce the accumulation of local positioning errors in constructing a global shape. Experimental work on material properties and dimensional scaling of the digital material leads to application in assembling functional structures. We propose that assembling digital material will be the future of 3-D free-form fabrication of functional materials.

We are developing a printer that builds functional, 3-D structures by reversible assembly of a discrete set of "digital materials." This approach uses the components rather than a control system to impose the spatial and functional constraints. Printing can be performed as a parallel rather than a linear process; the printing process is reversible for re-use of the pieces or error correction at any point in the object’s life. Error detection, error-reduction, and error-tolerance during assembly allow for reliable, high-throughput printing. We are presenting development approaches to such a printing device.

"Fab Lab" is an abbreviation for Fabrication Laboratory, a group of off-the-shelf, industrial-grade fabrication and electronics tools, wrapped in open-source software and programs. Fab Labs give users around the world the ability to locally conceptualize, design, develop, fabricate, and test almost anything. The engineering capability for design and fabrication at micron length and microsecond time-scales opens up numerous possibilities for innovative solutions to common problems. Since local communities foster this innovation, it can lead to sustainable solutions. As yet, high-end technological solutions have not been addressing problems faced on the local level; therefore, we believe Fab Labs will provide a thriving incubator for local micro-businesses.

Internet 0 is an experiment at networking at the ultra-lightweight scale. Instead of relying on the architectural notions of Internet 1 with its routing, servers, and layered network stacks, we are toying with very small, cheap, and simple ways to bring Internet Protocols all the way to the physical interface.

Luminoso is a tool that uses common sense and blending to "read between the lines" and better understand opinions and feedback expressed in free text such as customer reviews. It creates a semantic space from the ideas in a set of documents, including common-sense background information, and allows interactive exploration. This interface can be used to discover trends in a text corpus, such as free-text responses to a survey.

Within the next three years, another billion people will make regular use of cell phones, continuing the fastest adoption of a new technology in history. This will unleash a wave of entrepreneurship, collaboration, and wealth creation, thereby transforming this Next Billion into a powerful force in the global economy. By innovating applications and business models, and by testing them locally in the field, this initiative explores how mobile technologies can reduce friction in bottom of the pyramid markets across the developing world.

We are experimenting with technology for low-cost, large-area input devices, using change-source tomography on an electrically resistive medium.

This project will build comfortable technology and tools that a wearer can use to record their own speech in natural social settings, play it back, and learn about and reflect on how he sounds. The tools will include automated analysis of pitch patterns, loudness, and timing, to help highlight these non-linguistic features of speech. The toolkit can be used during, or in conjunction with, speech-language therapy for individuals on the autism spectrum to explore and reflect on natural social situations. It will also enable autism researches to collect speech data outside of laboratory settings to quantify progress with language and communication in natural social contexts.

Engineering in a limit of thermodynamic complexity, by compiling global optimizations into local dynamics in systems including analog logic, paintable computing, and quantum optics