MIT Chancellor Barnhart thanks Course 6’s SuperUROP students for their research and innovation contributions; praises SuperUROP’s quintessential MITness
 

As the 2015 SuperUROP graduates flowed into the year-end certificates presentation event on May 7 at the Media Lab Skyline Room, the class of over 90 students looked ready for celebrating — just as they had embarked this past year on their intense research collaboration with their faculty advisor and his/her research group — supported as research scholars by the generous backing of industry and private sponsors.

Senior Ava Soleimany says she has gained a “much nuanced understanding of what exactly research involves.” With an interest in the computational aspects of synthetic biology and its medical applications, Soleimany applied to be a SuperUROP in spring 2014 in Prof. Timothy Lu’s Synthetic Biology Group. She was able to extend her existing UROP to a SuperUROP. She notes “It allowed me to hit the ground running in the fall while many of my peers were still sorting out the details of their projects.” She intends to continue with work/research at the intersection of EECS and Biology, but learned that she wants to reach beyond synthetic biology. This summer she will do a research internship at Seven Bridges Genomics.

Following their group photo session on the deck of the Media Lab Skyline Room, the SuperUROP graduates gathered to hear MIT Chancellor Cynthia Barnhart warmly congratulate them. “I’d like you to know that your contributions to the research enterprise and the innovation ecosystem at MIT are recognized, are an important element and are deeply appreciated,” she told them.

Barnhart cited several SuperUROP graduates who had reflected on their experience. The words offered by 2013-14 SuperUROP student Ishwarya Ananthabhotla in the 2014-15 brochure resonated with her — and most likely with many of those in the room. “Being a SuperUROP showed me something that I think is easy to forget at MIT. Effort, patience, hard work, and persistence—often over a long period of time, such as an entire year— will leave you with a rewarding, successful learning experience.”

Junior Neerja Aggarwal, energized by her year of research under Professor Rajeev Ram, is looking forward to working as a “print process summer intern at FormLabs in Somerville, MA, with the goal of continuing her research when she returns to MIT working towards her MEng and potentially her PhD. Aggarwal particularly valued the weekly meetings her SuperUROP advisor provided. “We met every week and he offered guidance not only on my research project, but also on how to approach problems and thinking about my future.”

The SuperUROP includes a year-long class (6.UAR — Undergraduate Advanced Research) devoted to examining research in depth – from picking a topic to effective communications about a research project. The final classes included 90-second presentations from each student about his/her SuperUROP research topic. This year yielded research in the broad areas of robotics, circuits and devices, bioEECS, artificial intelligence, machine learning, system optimization, algorithms and image processing. Some projects built on previous SuperUROPs demonstrating the collaborative nature of the EECS research community. Many projects had working prototypes and simulations.

SuperUROP creator and director Anantha Chandrakasan was thrilled with the way the students rose to the challenges of the SuperUROP experience. As instructor of 6.UAR, with MIT’s Dean for Undergraduate Education Dennis Freeman, Chandrakasan noted about the presentations and the overall milestones, “The level was truly outstanding — and accessible by a broad audience. As with the previous class, we look forward to seeing many of these projects in premiere conferences and journals as well as watching our students move on to top PhD programs and industry positions.”

Seniors Erik Johnson and Ahmet Musabeyoglu, two Duke Energy scholars working under Prof. David Perreault, traveled to India to gather end-user input for their projects on rural electrification. With the goal of developing a scalable DC microgrid architecture with cost-effective, smart power converters, the two took their system prototype to Jarkhand India, successfully implementing their system in two different villages. “Seeing the happiness in the eyes of people who met electricity for the first time in their life was the most satisfactory result of my whole SuperUROP research,” said Musabeyoglu. The two plan to return to set up a second revision this summer.

The challenges that each SuperUROP encounters during her or his deep dive into research are not trivial and often deeply impactful. Working closely with graduate student supervisor Nathaniel Rouquet in the Lu Lab, Soleimany says that all of their experiments were failing during the fall. “Navigating that situation was a great learning experience,” she says, including “…how to persevere, motivate yourself, take chances, go out on a limb for a potential solution….” When their experiments started to work she notes, “it was even more gratifying/exciting, given all that we had dealt with.” That’s a reality of research,” she notes. “I would not have been able to understand that without SuperUROP.” She plans to finish up her part of this research, which will be published in a professional journal.

Junior Kaustav Gopinathan worked with Professors George Verghese and Collin Stultz on a project that explored and evaluated electrical techniques to help detect and quantify anemia, a disorder faced by almost 25% of the world’s population— significant because the populations carrying the highest burden of anemia often have the least access to hospital technologies. “I plan to extend the SuperUROP project further in the coming semesters,” says Gopinathan. “Looking ahead, I want to pursue an MD-PhD, and this SuperUROP project in BioEECS is exactly the kind of research I want to work on in the future.”

Chancellor Barnhart said in her remarks on May 7: “SuperUROP in its design is so quintessentially MIT, particularly because it builds on MIT’s tradition of mens et manus, and it is aligned with the Institute’s priorities of discovery, innovation, and making the world a better place.” She also noted the attractiveness of the SuperUROP to other departments and beyond, suggesting the possibility that other universities across the globe might adopt this model—just as they did with the original 1969 UROP program. At MIT, the AeroAstro Department will begin a SuperUROP in the upcoming 2016 academic year.

Professor Jaime Peraire, AeroAstro Department Head, is excited about piloting the program in his department. “We had anticipated 5-7 students signing up in the first year. When I was told 21 students had submitted proposals, with 12 AeroAstro faculty and research scientists agreeing to supervise their efforts, I was thrilled. Our students are excited to be working so closely with faculty and our industrial partners. The fact that we have been able to leverage so much support from industry in our very first year is tremendous. We’re eager to see what the future brings.”

As Barnhart posed to the SuperUROPs, faculty supervisors and SuperUROP staff, “What more could we ask for?”

A computer vision enabled technology developed by a team of EECS faculty Bill Freeman and Frédo Durand and their students is enabling a new way to identify structural defects in objects.  The group will report this latest work at the Conference on Computer Vision and Pattern Recognition in June. 

Read more in the May 21, 2015 MIT News Office article by Larry Hardesty titled "Gauging materials’ physical properties from video: “Visual microphone” technology could lead to noninvasive identification of objects’ structural defects," also posted below.

Last summer, MIT researchers published a paper describing an algorithm that can recover intelligible speech from the analysis of the minute vibrations of objects in video captured through soundproof glass.

In June, at the Conference on Computer Vision and Pattern Recognition, researchers from the same groups will describe how the technique can be adapted to infer material properties of physical objects, such as stiffness and weight, from video.

The technique could have application in the field of “nondestructive testing,” or determining materials’ physical properties without extracting samples from them or subjecting them to damaging physical tests. It might be possible, for instance, to identify structural defects in an airplane’s wing by analyzing video of its vibration during flight.

“One of the big contributions of this work is connecting techniques in computer vision to established theory on physical vibrations and to a whole body of work in nondestructive testing in civil engineering,” says Abe Davis, an MIT graduate student in electrical engineering and computer science who, together with fellow graduate student Katie Bouman, is first author on the paper. “We make this connection pretty explicitly in the paper, which is I think where a lot of the future potential lies, because it bridges these fields.”

Davis, Bouman, and their coauthors — their advisors, respectively, professors of computer science and engineering Fredo Durand and Bill Freeman; Justin Chen, a graduate student in civil and environmental engineering; and Michael Rubinstein, who completed his PhD with Freeman but is now at Google Research in Cambridge, Massachusetts — applied their technique to two different types of object. One was rods of fiberglass, wood, and metal; the other, fabrics draped over a line.

Vibrational fingerprint

In the case of the rods, they used a range of frequencies from a nearby loudspeaker to produce vibrations. And since the vibrational frequencies of stiff materials are high, they also used a high-speed camera — as they did in some of the visual-microphone work — to capture the video.

The fabrics, however, were flexible enough that the ordinary circulation of air in a closed room was enough to produce detectible vibrations. And the vibration rates were low enough that they could be measured using an ordinary digital camera.

Although its movement may be undetectable to the human eye, a vibrating object usually vibrates at several frequencies at the same time. A given object’s preferred frequencies, and the varying strength of its vibrations at those frequencies, produce a unique pattern, which a variation on the visual-microphone algorithm was able to extract.

The researchers then used machine learning to find correlations between those vibrational patterns and measurements of the objects’ material properties. The correlations they found provided good estimates of the elasticity of the bars and of the stiffness and weight per unit area of the fabrics. (Stiffness is a measure that factors in both elasticity, a property of a material, and the geometry of the object made from that material. The researchers knew the precise geometry of the rods but not that of fabrics, which varied somewhat in size.)

Moreover, aberrations or discontinuities in an object’s typical vibrational patterns could indicate a defect in its structure. Identifying those types of correlations is the subject of the researchers’ ongoing work.

EECS Celelbrates, the annual awards event to celebrate and recognize the outstanding accomplishments of the Electrical Engineering and Computer Science (EECS) Department faculty, students and staff was held at the Museum of Fine Arts on Saturday, May 16. Anantha Chandrakasan, EECS department head, introduced the program citing several outstanding faculty recognitions including Mildred Dresselhaus' being selected by President Obama for the Presidential Medal of Freedom as well as her being selected for the IEEE Medal of Honor. He also noted that a record five EECS faculty were elected to the National Academy of Engineering, and pointed to other news provided in the 2015 EECS Connector. The 2015 EECS Connector hardcopy edition, was made available at the event.

Chandrakasan was joined by associate department head David Perreault for faculty awards, then by co-education officers Harry Lee and Rob Miller for faculty teaching awards and by undergraduate education officer Albert Meyer for student teaching and other recognition awards. Prof. Devavrat Shah announced the student awards followed by several student group awards to faculty for their service.

See the slides below with photos taken at the event and listing each award and recipient.

EECS graduate students Andrew Spielberg and Stuart Baker and postdoc Mehmet Dogar with EECS Professor and Director of the Computer Science and Artificial Intelligence Lab (CSAIL) Daniela Rus have developed a new algorithm that significantly reduces the time it takes for several robots to plan and execute a task. Their work was cited this week by the Institute for Electrical and Electronics Engineers’ International Conference on Robotics and Automation with two best-paper awards. [Photo above: MIT researchers tested the viability of their algorithm by using it to guide a crew of three robots in the assembly of a chair.  Photo credit: Dominick Reuter/ MIT News]

Read more in the May 27, 2015 MIT News Office article by Larry Hardesty titled "Helping robots put it all together - New algorithm lets autonomous robots divvy up assembly tasks on the fly," also posted below.

Today’s industrial robots are remarkably efficient — as long as they’re in a controlled environment where everything is exactly where they expect it to be. But put them in an unfamiliar setting, where they have to think for themselves, and their efficiency plummets. And the difficulty of on-the-fly motion planning increases exponentially with the number of robots involved. For even a simple collaborative task, a team of, say, three autonomous robots might have to think for several hours to come up with a plan of attack.

This week, at the Institute for Electrical and Electronics Engineers’ International Conference on Robotics and Automation, a group of MIT researchers were nominated for two best-paper awards for a new algorithm that can significantly reduce robot teams’ planning time. The plan the algorithm produces may not be perfectly efficient, but in many cases, the savings in planning time will more than offset the added execution time.

The researchers also tested the viability of their algorithm by using it to guide a crew of three robots in the assembly of a chair.

“We’re really excited about the idea of using robots in more extensive ways in manufacturing,” says Daniela Rus, the Andrew and Erna Viterbi Professor in MIT’s Department of Electrical Engineering and Computer Science, whose group developed the new algorithm. “For this, we need robots that can figure things out for themselves more than current robots do. We see this algorithm as a step in that direction.”

Rus is joined on the paper by three researchers in her lab — first author Mehmet Dogar, a postdoc, and Andrew Spielberg and Stuart Baker, both graduate students in electrical engineering and computer science.

Grasping consequences

The problem the researchers address is one in which a group of robots must perform an assembly operation that has a series of discrete steps, some of which require multirobot collaboration. At the outset, none of the robots knows which parts of the operation it will be assigned: Everything’s determined on the fly.

Computationally, the problem is already complex enough, given that at any stage of the operation, any of the robots could perform any of the actions, and during the collaborative phases, they have to avoid colliding with each other. But what makes planning really time-consuming is determining the optimal way for each robot to grasp each object it’s manipulating, so that it can successfully complete not only the immediate task, but also those that follow it.

“Sometimes, the grasp configuration may be valid for the current step but problematic for the next step because another robot or sensor is needed,” Rus says. “The current grasping formation may not allow room for a new robot or sensor to join the team. So our solution considers a multiple-step assembly operation and optimizes how the robots place themselves in a way that takes into account the entire process, not just the current step.”

The key to the researchers’ algorithm is that it defers its most difficult decisions about grasp position until it’s made all the easier ones. That way, it can be interrupted at any time, and it will still have a workable assembly plan. If it hasn’t had time to compute the optimal solution, the robots may on occasion have to drop and regrasp the objects they’re holding. But in many cases, the extra time that takes will be trivial compared to the time required to compute a comprehensive solution.

Principled procrastination

The algorithm begins by devising a plan that completely ignores the grasping problem. This is the equivalent of a plan in which all the robots would drop everything after every stage of the assembly operation, then approach the next stage as if it were a freestanding task.

Then the algorithm considers the transition from one stage of the operation to the next from the perspective of a single robot and a single part of the object being assembled. If it can find a grasp position for that robot and that part that will work in both stages of the operation, but which won’t require any modification of any of the other robots’ behavior, it will add that grasp to the plan. Otherwise, it postpones its decision.

Once it’s handled all the easy grasp decisions, it revisits the ones it’s postponed. Now, it broadens its scope slightly, revising the behavior of one or two other robots at one or two points in the operation, if necessary, to effect a smooth transition between stages. But again, if even that expanded scope proves too limited, it defers its decision.

If the algorithm were permitted to run to completion, its last few grasp decisions might require the modification of every robot’s behavior at every step of the assembly process, which can be a hugely complex task. It will often be more efficient to just let the robots drop what they’re holding a few times rather than to compute the optimal solution.

In addition to their experiments with real robots, the researchers also ran a host of simulations involving more complex assembly operations. In some, they found that their algorithm could, in minutes, produce a workable plan that involved just a few drops, where the optimal solution took hours to compute. In others, the optimal solution was intractable — it would have taken millennia to compute. But their algorithm could still produce a workable plan.

“With an elegant heuristic approach to a complex planning problem, Rus’s group has shown an important step forward in multirobot cooperation by demonstrating how three mobile arms can figure out how to assemble a chair,” says Bradley Nelson, the Professor of Robotics and Intelligent Systems at Swiss Federal Institute of Technology in Zurich. “My biggest concern about their work is that it will ruin one of the things I like most about Ikea furniture: assembling it myself at home.”

Sangeeta Bhatia, member of the EECS faculty, of MIT’s Koch Institute for Integrative Cancer Research and the Institute for Medical Engineering and Science and has teamed with researchers at the University of California at San Diego to create a new way to detect cancer that has spread to the liver by using probiotics — providing a way for earlier detection of cancers such as colon and pancreatic that typcially metastasize to the liver. 

Read more in the May 27, 2015 MIT News Office article by Anne Trafton titled "Diagnosing cancer with help from bacteria. Engineered probiotics can detect tumors in the liver," also posted below.

Engineers at MIT and the University of California at San Diego (UCSD) have devised a new way to detect cancer that has spread to the liver, by enlisting help from probiotics — beneficial bacteria similar to those found in yogurt.

Many types of cancer, including colon and pancreatic, tend to metastasize to the liver. The earlier doctors can find these tumors, the more likely that they can successfully treat them.

“There are interventions, like local surgery or local ablation, that physicians can perform if the spread of disease in the liver is confined, and because the liver can regenerate, these interventions are tolerated. New data are showing that those patients may have a higher survival rate, so there’s a particular need for detecting early metastasis in the liver,” says Sangeeta Bhatia, the John and Dorothy Wilson Professor of Health Sciences and Electrical Engineering and Computer Science at MIT.

Using a harmless strain of E. coli that colonizes the liver, the researchers programmed the bacteria to produce a luminescent signal that can be detected with a simple urine test. Bhatia and Jeff Hasty, a professor of biology at UCSD, are the senior authors of a paper describing the new approach this week in the journal Science Translational Medicine. Lead authors are MIT postdoc Tal Danino and UCSD postdoc Arthur Prindle.

Microbial help

Previous studies had shown that bacteria can penetrate and grow in the tumor microenvironment, where there are lots of nutrients and the body’s immune system is compromised. Because of this, scientists have been trying for many years to develop bacteria as a possible vehicle for cancer treatment.

The MIT and UCSD researchers began exploring this idea a few years ago, but soon expanded their efforts to include the concept of creating a bacterial diagnostic.

To turn bacteria into diagnostic devices, the researchers engineered the cells to express the gene for a naturally occurring enzyme called lacZ that cleaves lactose into glucose and galactose. In this case, lacZ acts on a molecule injected into the mice, consisting of galactose linked to luciferin, a luminescent protein naturally produced by fireflies. Luciferin is cleaved from galactose and excreted in the urine, where it can easily be detected using a common laboratory test.

At first, the researchers were interested in developing these bacteria for injection into patients, but then decided to investigate the possibility of delivering the bacteria orally, just like the probiotic bacteria found in yogurt. To achieve that, they integrated their diagnostic circuits into a harmless strain of E. coli called Nissle 1917, which is marketed as a promoter of gastrointestinal health.

In tests with mice, the researchers found that orally delivered bacteria do not accumulate in tumors all over the body, but they do predictably zero in on liver tumors because the hepatic portal vein carries them from the digestive tract to the liver.

“We realized that if we gave a probiotic, we weren’t going to be able to get bacteria concentrations high enough to colonize the tumors all over the body, but we hypothesized that if we had tumors in the liver they would get the highest dose from an oral delivery,” says Bhatia, who is a member of MIT’s Koch Institute for Integrative Cancer Research and Institute for Medical Engineering and Science.

This allowed the team to develop a diagnostic specialized for liver tumors. In tests in mice with colon cancer that has spread to the liver, the probiotic bacteria colonized nearly 90 percent of the metastatic tumors.

In the mouse experiments, animals that were given the engineered bacteria did not exhibit any harmful side effects.

More sensitive detection

The researchers focused on the liver not only because it is a natural target for these bacteria, but also because the liver is hard to image with conventional imaging techniques like CT scanning or magnetic resonance imaging (MRI), making it difficult to diagnose metastatic tumors there.

With the new system, the researchers can detect liver tumors larger than about one cubic millimeter, offering more sensitivity than existing imaging methods. This kind of diagnostic could be most useful for monitoring patients after they have had a colon tumor removed because they are at risk for recurrence in the liver, Bhatia says.

Andrea Califano, a professor of biological sciences at Columbia University, says the study is “seminal and thought-provoking in terms of clearing a new path for investigating what can be done for early detection of cancer,” adding that the therapeutic possibilities are also intriguing.

“These bacteria could be engineered to cause genetic disruption of cancer cell function, deliver drugs, or reactivate the immune system,” says Califano, who was not involved in the research.

The MIT team is now pursuing the idea of using probiotic bacteria to treat cancer, as well as for diagnosing it.

The research was funded by the Ludwig Center for Molecular Oncology at MIT, a Prof. Amar G. Bose Research Grant, the National Institutes of Health through the San Diego Center for Systems Biology, and the Koch Institute Support Grant from the National Cancer Institute.

Two EECS juniors, Kaustav Gopinathan and Margaret Guo, have been selected as Goldwater Scholars for 2015-16. Four MIT students, honored Goldwater Scholars for their academic achievements, were selected from a field of 1,206 candidates nominated by university faculty nationwide. [Photo: 2015 MIT Goldwater Scholars: (l-r) Julia Page, Felipe Hernandez, Kaustav Gopinathan, and Margaret Guo. Photo: Lillie Paquette/School of Engineering]

Read more in the May 29, 2015 MIT News Office article by Leda Zimmerman titled "Meet the 2015 Goldwater Scholars - Four MIT students honored for their academic achievements," also posted below.

Four MIT juniors have been named recipients of Barry Goldwater Scholarship Awards for 2015-16. They were selected on the basis of academic merit from a field of 1,206 candidates nominated by university faculty nationwide. This year’s Goldwater Scholarship recipients are Kaustav A. Gopinathan, Margaret G. Guo, Felipe Hernandez, and Julia E. Page.

Gopinathan, majoring in electrical engineering and computer science (EECS), “is the most talented undergraduate student I have ever encountered … destined to be a scholar of the highest quality and I look forward to seeing his name in lights,” wrote one faculty member in his recommendation, adding “I typically do not write words of praise liberally.” Gopinathan, who has conducted research to develop a low-cost medical device for diagnosing anemia, and a signal processing technique for identifying apnea in newborns, intends to acquire both an MD and PhD.

Guo, a double major in EECS and biological engineering, hopes to perform research to increase understanding of biological systems, focusing “on engineering tractable models … for the purposes of supporting clinical decision making or improving biomedical systems and devices.” She got an early start on such research. In an internship with Medtronics, Guo helped to develop a new generation pacemaker, and in the lab of Linda Griffith, the School of Engineering Professor of Teaching Innovation and a professor of biological and mechanical engineering, Guo worked on image and statistical analysis tools used in an organ model for endometriosis.

Hernandez, majoring in mathematics, intends to pursue a PhD in this field and advance understanding between analysis, combinatorics, geometric measure theory, and materials science. One faculty advisor wrote that “what is really amazing is his ability to learn independently, and I believe that he is on track to become a first-rate research mathematician and scientist.”

Page, majoring in chemistry, plans to conduct research at the intersection of chemistry and medicine, focusing on diseases and the drugs used to treat them at the molecular level. One of Page’s recommendations concluded: “She is one of the most talented students I have met in more than two decades on the faculty at MIT. Julia has outstanding potential for leadership in a research career.” Page’s interest in biomedical research is motivated in part by her experience shadowing a radiation oncologist and engaging with cancer patients. Says Page, “I would like to have a career that combines research with some patient care.”

The Barry Goldwater Scholarship and Excellence in Education Program was established by Congress in 1986 to honor Senator Barry Goldwater, who served for 30 years in the U.S. Senate. The program is designed to encourage outstanding students to pursue careers in math, the natural sciences, and engineering. Recipients will receive stipends covering the cost of tuition, fees, books, and room and board up to a maximum of $7,500 per year.

Jing Kong, professor of electrical engineering in the EECS Department at MIT and principal investigator with the Microsystems Technology Laboratories and the Research Lab of Electronics has worked with MIT students and Evelyn Wang, professor in MIT's Mechanical Engineering Department to create a graphene coating for power plant condensers — a step that will improve power plant efficiency 2 to 3 percent and ultimately making a significant dent in global carbon emissions. [Image: An uncoated copper condenser tube (top left) is shown next to a similar tube coated with graphene (top right). When exposed to water vapor at 100 degrees Celsius, the uncoated tube produces an inefficient water film (bottom left), while the coated shows the more desirable dropwise condensation (bottom right). Courtesy of the researchers and MIT News.]

Read more in the May 29, 2015 MIT News Office article by David L. Chandler titled "Thin coating on condensers could make power plants more efficient Graphene layer one atom thick could quadruple rate of condensation heat transfer in generating plants," also posted below.

Most of the world’s electricity-producing power plants — whether powered by coal, natural gas, or nuclear fission — make electricity by generating steam that turns a turbine. That steam then is condensed back to water, and the cycle begins again.

But the condensers that collect the steam are quite inefficient, and improving them could make a big difference in overall power plant efficiency.

Now, a team of researchers at MIT has developed a way of coating these condenser surfaces with a layer of graphene, just one atom thick, and found that this can improve the rate of heat transfer by a factor of four — and potentially even more than that, with further work. And unlike polymer coatings, the graphene coatings have proven to be highly durable in laboratory tests.

The findings are reported in the journal Nano Letters by MIT graduate student Daniel Preston, professors Evelyn Wang and Jing Kong, and two others. The improvement in condenser heat transfer, which is just one step in the power-production cycle, could lead to an overall improvement in power plant efficiency of 2 to 3 percent based on figures from the Electric Power Research Institute, Preston says — enough to make a significant dent in global carbon emissions, since such plants represent the vast majority of the world’s electricity generation. “That translates into millions of dollars per power plant per year,” he explains.

There are two basic ways in which the condensers — which may take the form of coiled metal tubes, often made of copper — interact with the flow of steam. In some cases, the steam condenses to form a thin sheet of water that coats the surface; in others it forms water droplets that are pulled from the surface by gravity.

When the steam forms a film, Preston explains, that impedes heat transfer — and thus reduces the efficiency — of condensation. So the goal of much research has been to enhance droplet formation on these surfaces by making them water-repelling.

Often this has been accomplished using polymer coatings, but these tend to degrade rapidly in the high heat and humidity of a power plant. And when the coatings are made thicker to reduce that degradation, the coatings themselves impede heat transfer.

“We thought graphene could be useful,” Preston says, “since we know it is hydrophobic by nature.” So he and his colleagues decided to test both graphene’s ability to shed water, and its durability, under typical power plant conditions — an environment of pure water vapor at 100 degrees Celsius.

They found that the single-atom-thick coating of graphene did indeed improve heat transfer fourfold compared with surfaces where the condensate forms sheets of water, such as bare metals. Further calculations showed that optimizing temperature differences could boost this improvement to 5 to 7 times. The researchers also showed that after two full weeks under such conditions, there was no measurable degradation in the graphene’s performance.

By comparison, similar tests using a common water-repelling coating showed that the coating began to degrade within just three hours, Preston says, and failed completely within 12 hours.

Because the process used to coat the graphene on the copper surface — called chemical vapor deposition — has been tested extensively, the new method could be ready for testing under real-world conditions “in as little as a year,” Preston says. And the process should be easily scalable to power plant-sized condenser coils.

“This work is extremely significant because, to my knowledge, it is the first report of durable dropwise condensation with a single-layer surface coating,” says Jonathan Boreyko, an assistant professor of biomedical engineering and mechanics at Virginia Tech who has studied condensation on superhydrophobic surface. “These findings are somewhat surprising and very exciting.”

Boreyko, who was not involved in the research, adds that this method, if proven through further testing, “could significantly improve the efficiency of power plants and other systems that utilize condensers.”

The research team also included MIT postdoc Daniela Mafra and former postdoc Nenad Miljkovic, who is now an assistant professor at the University of Illinois at Urbana-Champaign. The work was supported by the Office of Naval Research and the National Science Foundation.

It has been nearly four months since 70 MIT undergraduate and graduate students worked in teams for a single day at the Assistive Technologies Hackathon (ATHack 2015) to create a prototype to improve an aspect in the daily routine of their client, a person living with a disability. ATHack 2015 was dedicated to honor the late Seth Teller, who had inspired MIT students with the class he created in the fall of 2011 titled 6.811, Principles and Practices in Assistive Technologies, also known as PPAT. Held every fall since, students had started the ATHack in 2014 as a way to extend this exposure to assistive technologies in the spring term.

The beauty of this event is that in some cases, work initiated during the ATHack does not end there. Here is the story of one team, the winning team, as they continue to bring their prototype to reality.

By day’s end on Feb. 28, at the MIT Beaver Works in Technology Square in Cambrige, three teams and their resulting prototypes were announced as winners of the ATHack 2015 top three prizes. Team Adriana, named as each team traditionally has been — for the first name of their client — won first place for the design of what they call “Puffin” — a sip and puff joy-stick controller designed to make it possible for their client or anyone living with cerebral palsy or without upper body dexterity, to use a smart phone, tablet or laptop. [Read the March 16, 2015 MIT News Office article.] 

Team Adriana members Ned Burnell, PhD in MechE, Esther Jang, MEng student in EECS, Shirlene Liew, Master’s student in Systems Design and Management, and Kate Tatar, MechE undergraduate, gained that name and assignment at a pre-event dinner held roughly two weeks before the hack. Project ideas had already been worked out in advance by the clients and were then presented by the ATHack event directors Ishwarya Ananthabhotla, Jaya Narain, and Abigail Klein. These three were aided by event organizers William Li, Emma Nelson, Sneha Lingam, Jennifer Tylock, Dhruv Jain and Beth Rosen-Filardo.

In advance of the ATHack, each team was matched to a client — resulting in 17 teams to cover the needs of 14 clients. Several clients were given more than one team to work on multiple assistive technologies. The needs for an assistive technology that were presented to the Hack teams arose from living with disabilities ranging from cerebral palsy, paraplegia, diabetes, neuropathic facial pain and hearing and vision impairments.

Jang notes about this pre-event stretch: “A lot of advance preparation happened after the matching.” The group worked with Adriana during the post dinner week, creating sketches and ideas based on pictures she sent and details about her wheelchair setup. Based on this work, the team came up with an overall design and the different functional elements they would need: a structural element to hold the mouse, the physical joystick with pressure sensing built in, and the microcontroller plus Bluetooth modules to receive the signals from the joystick and send to the controlled device. This was a start.

Burnell and his team members were impressed with Adriana when, at the dinner as the organizers presented her project and didn’t quite have the details down, she was handed the microphone to provide the description. “Adriana seemed like a great person to work with,” Burnell notes. “When we first approached Adriana, Kate and Esther mostly carried the conversation, while I looked parts up online and asked implementation-specific questions,” he recalls. Within ten minutes, the group had finalized their materials for the ATHack and Adriana selected them as her team.

Adriana Mallozzi says she’s been addicted to technology since age 7, when she was able to use a computer — her first taste of independence, she notes. Living with cerebral palsy from birth, she has been associated since age 1 with Easter Seals, the nonprofit, charitable organization that assists children and adults with disabilities. ATHack 2015 was her first hackathon experience. “I was in heaven that day and was so glad that the team won,” she says, admitting that she is still on a high after the experience.

The actual ATHack became an all out attempt by the team to meet the challenges and snags to complete the things that weren’t planned beforehand. “Though the final prototype was held together with tape and superglue — instead of screws or expoxy,” says Jang, “and a lot of epidermal cells lost in the process — the team won first place."

Besides the team’s persistence and ingenuity, the real glue was the presence of client Adriana, who worked by her team all day. Burnell says that Adriana helped test things like the pressure sensor (the threshold for sips and puffs had to be reasonable for her breaths), the length/shape of the joystick mouse and the positioning of the structural elements so that she could use them comfortably. “Her suggestions such as melting the plastic mouthpiece to be a more comfortable shape for her all made the process work.”

In retrospect, Burnell says his closest experience with client-based design was in high school when he helped people with age-related disabilities configure their computers to be more pleasant and less frustrating to use. He says that it was helpful for Team Adriana members to think out loud about how to do what she outlined — helping her to communicate what the actual need was. “We don't want to just listen to her and then run away to work on it any more than she wants to be the oracle of what is needed with no input into how it's done,” Burnell noted. “Design is a conversation,” he continued.

Since then, Jang and Burnell, with Adriana’s assistance, have applied for small grants around MIT so they can make half a dozen Puffins for testing this summer. They have already received an MIT Public Service Center (PSC) LEAP grant for $1000. There is a good chance that through Easter Seals and Adriana’s connections, the Puffin will become known to people who will find it useful — providing real-time testing and application in the ways for which it has been so carefully designed. [See the Puffin website: http://puffinsip.com/]

Enabling neighbors in remote areas of the world that lack electricity to inexpensively buy power from a local solar panel owner, essentially creating their own microgrid, may sound utopian, but it is being developed by EECS faculty members Rajeev Ram and David Perreault as well as MIT doctoral students Wardah Inam (EECS) and David Stawser (MechE) and research associate Reja Amatya, SuperUROP students Ahmet Musabeyoglu and Erik Johnson, and graduate student Varun Mehra. [Photo: The key element of the new microgrid system is the power management unit, seen at the center in this rendering. Various devices — shown here as a fan, a light, and a cellphone charger — can be plugged directly into the unit, along with lines to supply power to other houses. The unit manages the power coming in from solar panels (shown at top), and sends the power either to the devices or, if not needed immediately, to storage batteries (right of center). The display panel can provide information about usage.  Courtesy of the MIT team/MIT News]

Read more in the June 1, 2015 MIT News Office article by David L. Chandler titled "Bringing microgrids to rural villages - MIT team, working with villagers in India, designs peer-to-peer system to enable local power sharing," also posted below

An estimated 1.3 billion people around the world lack access to electricity, and as a result spend scarce resources on kerosene and other fuels for lighting. Now MIT researchers have developed a system to enable those in rural villages who can afford solar panels to share power with their neighbors, providing both income for the owners and much-needed power for the neighbors.

The key to the system, developed over two years of research and numerous trips to India, lies in a simple device the team developed that is smaller than a shoebox. The power management unit (PMU) performs a variety of tasks, regulating how electricity from solar panels or other sources gets directed to immediate uses — such as powering lights and cellphones — or to batteries for later use. At the same time, the PMU monitors how much power is going to each user, providing a record that can be used for billing without a need for individual meters.

MIT doctoral students Wardah Inam and Daniel Strawser, under the guidance of electrical engineering professors Rajeev Ram and David Perreault, will head to India next week, along with several other team members, to spend the summer doing field tests of the system. Along the way, they will stop off in Seoul, South Korea, to present an account of their work at the International Conference on Power Electronics.

Earning extra revenue

Test installations will take place in two villages in the Jamshedpur area in northeastern India — one of which has no outside power source at all, and one of which is connected to the grid, but gets only intermittent access, averaging two to three hours of electricity a day. Some people in these villages “have never interacted with this kind of technology before,” Inam says of the findings from their previous trips, where they met with village leaders and residents to discuss their needs.

A few of the villages’ houses already have small, simple solar-power systems set up to power a few low-power LED lights and charge cellphones. These early solar installations, Inam explains, will now provide their owners with an opportunity to earn revenue by selling excess power to neighbors who lack any source of electricity.

Unlike typical solar installations in the area — where every lamp, fan, or charger is hard-wired to the system — the new MIT-designed systems will allow for flexibility in adding or removing lights or other devices; adding extra power sources, including more solar panels or other sources such as diesel generators; and adding connections for additional users over time.

While most of the world’s electric grid systems use AC (alternating current), the new MIT systems operate entirely on DC (direct current), which greatly simplifies setup, lowers costs, and is safer for users to operate. Since the typical uses of electricity in these settings — lighting, charging phones, and powering fans — are all either inherently DC systems or can easily be converted to DC, and solar panels by their nature produce DC, this simplification eliminates the need for multiple devices to convert DC to AC, and back again. And because it is designed to operate at less than 50 volts, the systems are not capable of delivering life-threatening shocks even if wiring gets damaged and people are accidentally exposed to bare wires, the researchers say.

Designing for local needs

Rather than bringing in a system designed by outsiders, Inam says, the idea is for locals to develop a system to meet their own specific needs and preferences. “We want to empower the people to build a grid,” she explains. Already, some in these villages can afford to pay for solar installations at their homes, or can obtain a loan to finance it. But being able to sell some of that power to other nearby homes through a “microgrid” could enable users to buy larger systems, with some of the cost defrayed by power sales.

Meanwhile, those neighbors, without having to pay any upfront installation costs, would get the benefit of power for lights and charging for an estimated cost of $2 to $5 a month — “less than what they pay now for lighting, using kerosene or candles,” Inam says. “For the same amount, they’ll get better, safer lighting, as well as other services.”

Phone charging is one high-demand service that would be met by such a system. Cellphone service is widespread in India and other developing nations, but people often have to travel to a nearby city, or pay steep prices locally, to charge their phones. In addition, the solar power would enable them to use electric fans, a serious need in this very hot region. “A lot of solar home systems now don’t have that capability,” Inam says.

Another advantage of the microgrid approach is that it allows for the use of larger solar panels than many single-home systems could justify. This is crucial because most solar-panel mass-production worldwide is geared toward 250-watt panels, but single-home systems are often designed for 50-watt panels, where the cost per watt is much higher.

“We think this is really scalable,” Strawser says, adding that it could be profitable at many stages — for companies that provide the components, local workers who do the installations, solar panel owners who get revenue from the power, and end users who get power that not only facilitates reading and working after dark, but can also power water pumps for irrigation and other uses.

“It’s a bottom-up approach,” he says, in which local users — who have already contributed to the design choices through their interactions with the team — get to decide what gets installed where and when.

Saurabh H. Mehta, an energy professional in India who is not associated with this project, says that in seven years of looking at solar installations in that country, it has always bothered him “how all of the expensive power produced is not fully utilized. With this technology this certainly can change, as the owners of the system can easily supply surplus power to their neighbors. This can ultimately make the system more affordable and thus should encourage more uptake of off-grid renewable energy systems.”

The team also includes research associate Reja Amatya, undergraduates (and SuperUROP students) Ahmet Musabeyoglu and Erik Johnson, and graduate student Varun Mehra, who will all be working on the project in India this summer. Graduate students Szymon Sidor and Victor Lesniewski and former visiting professor Khurram Afridi also contributed to the team. The work was supported by the MIT Tata Center for Technology and Design.

Luis Fernando Velasquez-Garcia, principal research scientist in MIT's Microsystems Technology Laboratories (MTL) has led work to develop a new efficient and productive way to manufacture nanofibers througth electrospinning, mimicing dot-matrix printing with potential for many applications such as solar cell production.  [Image: A scanning electron micrograph of the new microfiber emitters, showing the arrays of rectangular columns etched into their sides. Courtesy of the researchers]

Read moe in the June 4, 2015 MIT News Office article by Larry Hardesty titled "Unlocking nanofibers’ potential - Prototype boosts production of versatile fibers fourfold, while cutting energy consumption by 92 percent," also posted below.

Nanofibers — polymer filaments only a couple of hundred nanometers in diameter — have a huge range of potential applications, from solar cells to water filtration to fuel cells. But so far, their high cost of manufacture has relegated them to just a few niche industries.

In the latest issue of the journal Nanotechnology, MIT researchers describe a new technique for producing nanofibers that increases the rate of production fourfold while reducing energy consumption by more than 90 percent, holding out the prospect of cheap, efficient nanofiber production.

“We have demonstrated a systematic way to produce nanofibers through electrospinning that surpasses the state of the art,” says Luis Fernando Velásquez-García, a principal research scientist in MIT’s Microsystems Technology Laboratories, who led the new work. “But the way that it’s done opens a very interesting possibility. Our group and many other groups are working to push 3-D printing further, to make it possible to print components that transduce, that actuate, that exchange energy between different domains, like solar to electrical or mechanical. We have something that naturally fits into that picture. We have an array of emitters that can be thought of as a dot-matrix printer, where you would be able to individually control each emitter to print deposits of nanofibers.”

Tangled tale

Nanofibers are useful for any application that benefits from a high ratio of surface area to volume — solar cells, for instance, which try to maximize exposure to sunlight, or fuel cell electrodes, which catalyze reactions at their surfaces. Nanofibers can also yield materials that are permeable only at very small scales, like water filters, or that are remarkably tough for their weight, like body armor.

The standard technique for manufacturing nanofibers is called electrospinning, and it comes in two varieties. In the first, a polymer solution is pumped through a small nozzle, and then a strong electric field stretches it out. The process is slow, however, and the number of nozzles per unit area is limited by the size of the pump hydraulics.

The other approach is to apply a voltage between a rotating drum covered by metal cones and a collector electrode. The cones are dipped in a polymer solution, and the electric field causes the solution to travel to the top of the cones, where it’s emitted toward the electrode as a fiber. That approach is erratic, however, and produces fibers of uneven lengths; it also requires voltages as high as 100,000 volts.

Thinking small

Velásquez-García and his co-authors — Philip Ponce de Leon, a former master’s student in mechanical engineering; Frances Hill, a former postdoc in Velásquez-García’s group who’s now at KLA-Tencor; and Eric Heubel, a current postdoc — adapt the second approach, but on a much smaller scale, using techniques common in the manufacture of microelectromechanical systems to produce dense arrays of tiny emitters. The emitters’ small size reduces the voltage necessary to drive them and allows more of them to be packed together, increasing production rate.

At the same time, a nubbly texture etched into the emitters’ sides regulates the rate at which fluid flows toward their tips, yielding uniform fibers even at high manufacturing rates. “We did all kinds of experiments, and all of them show that the emission is uniform,” Velásquez-García says.

To build their emitters, Velásquez-García and his colleagues use a technique called deep reactive-ion etching. On either face of a silicon wafer, they etch dense arrays of tiny rectangular columns — tens of micrometers across — which will regulate the flow of fluid up the sides of the emitters. Then they cut sawtooth patterns out of the wafer. The sawteeth are mounted vertically, and their bases are immersed in a solution of deionized water, ethanol, and a dissolved polymer. When an electrode is mounted opposite the sawteeth and a voltage applied between them, the water-ethanol mixture streams upward, dragging chains of polymer with it. The water and ethanol quickly dissolve, leaving a tangle of polymer filaments opposite each emitter, on the electrode.

The researchers were able to pack 225 emitters, several millimeters long, on a square chip about 35 millimeters on a side. At the relatively low voltage of 8,000 volts, that device yielded four times as much fiber per unit area as the best commercial electrospinning devices.

The work is “an elegant and creative way of demonstrating the strong capability of traditional MEMS [microelectromechanical-systems] fabrication processes toward parallel nanomanufacturing,” says Reza Ghodssi, a professor of electrical engineering at the University of Maryland. Relative to other approaches, he adds, there is “an increased potential to scale it up while maintaining the integrity and accuracy by which the processing method is applied.”

We are very proud to congratulate the 101 EECS Students who were recognized Thursday, June 4th during the MIT Investiture of Doctoral Hoods ceremony. 

An EECS Hooding Celebration was held following the ceremony where all in attendance shared in celebrating our students accomplishments. 

Our newest PhD's are pictured here along with EECS Department Head, Anantha Chandrakasan and EECS Graduate Officer, Leslie Kolodziejski.

Well done one and all!  CONGRATULATIONS!!!

Ethernet inventor and 3Com founder Robert Metcalfe’68 will become a visiting innovation fellow at MIT for one week a month during the 2015-16 academic year, engaging in entrepreneurship activities at the MIT Innovation Initiative and in the Department of Electrical Engineering and Computer Science (EECS). 

Read more in the June 8, 2015 article by Kelly Courtney, MIT Innovation Initiative, titled "Metcalfe to serve as visiting innovation fellow for 2015-16 academic year - Ethernet inventor and 3Com founder will shape Start6, EECS’s innovation and entrepreneurship workshop," also posted below.

As visiting innovation fellow, Robert Metcalfe succeeds former Massachusetts Gov. Deval Patrick, who held the post this spring.

Metcalfe, a professor of innovation in the Cocknell School of Engineering at the University of Texas at Austin, will engage with MIT students and mentor them in startup activities; participate in campus innovation events; hold office hours; participate in innovation roundtables; and serve as an advisor to the Lab for Innovation Science and Policy. He will play a significant role in helping to shape Start6, an EECS-launched workshop that immerses students and postdocs in innovation and entrepreneurship, providing them with the resources to translate their passion into needed technology solutions.

Start6 exposes graduate and undergraduate students, as well as postdocs, to various entrepreneurial paths. By meeting successful entrepreneurs and leaders in venture capital, students have the opportunity to broadly engage innovation resources at MIT and the supporting entrepreneurial ecosystem, so they can shape their own success.

“Innovation drives the virtuous cycle of freedom and prosperity,” Metcalfe says. “Startups out of research universities have proven to be among the most effective ways of innovating — startups built the Internet. It is exciting to have this opportunity to link the startup ecosystems of Austin and Boston.”

Metcalfe is a life member emeritus of the MIT Corporation — the Institute’s board of trustees — and an Internet pioneer: While a computer scientist at Xerox’s Palo Alto Research Center, he was the lead inventor of Ethernet. He then became an entrepreneur, founding and growing the multibillion-dollar networking company 3Com, now part of Hewlett-Packard. Metcalfe is a member of the National Academy of Engineering and, in 2005, received the National Medal of Technology.

“I have known Bob Metcalfe for more than 15 years, from when I served as associate department head in EECS,” MIT President L. Rafael Reif says. “Bob’s influence in today’s Internet-centered world is hard to overstate. He has a discerning mind that engages creatively on a very wide range of topics, and he asks the kind of tough questions that push you to imagine something entirely new. I have learned a great deal from Bob, and I believe others in our community will also enjoy and learn from his daring way of thinking.”

Innovation Initiative co-director Vladimir Bulović, the Fariborz Maseeh Professor of Emerging Technology and associate dean for innovation in MIT’s School of Engineering, says he looks forward to harnessing Metcalfe’s expertise in technology and entrepreneurship to further educate the next generation of global innovators.

“Since its founding, MIT has aspired to educate students at the intersection of theoretical and practical skills,” Bulović says. “Bob is an innovator who excelled in translating fundamental ideas into broad impact by connecting every aspect of the innovation ecosystem. His vast experience will bring enormous value to our students.”

For eight years, Metcalfe was publisher of “IDG InfoWorld,” writing a column that was read weekly by hundreds of thousands of information technologists. He has authored books including “Packet Communication” (PN, 1973) and “Internet Collapses and Other InfoWorld Punditry” (IDG Books, 2000). He is an emeritus partner of the Massachusetts-based venture capital firm Polaris Venture Partners.

Metcalfe earned two bachelor’s degrees, in electrical engineering and in industrial management, from MIT in 1969. He then went on to earn a master’s degree in applied mathematics and a Ph.D. in computer science from Harvard University.

The MIT DARPA Robotics Challenge Team led by Professor Rus Tedrake reached new heights in the June 5-6 international DARPA Robotics Challenge in Pomona California, as they nimbly programmed their Atlas robot to perform a wide range of tasks in one hour. The goal of the event was to develop mobile robots to perform useful tasks in disaster-relief situations — in response to the 2011 Fukushima nuclear disaster. [Photo of MIT DRC Team by Jason Dorfman/CSAIL.  Video, below, by Lillie Paquette/MIT School of Engineering]

Read more in the June 8, 2015 MIT News Office article by Adam Conner-Simons titled "MIT team places sixth at international DARPA Robotics Challenge CSAIL team just misses winning the grand prize after programming a 400-lb humanoid robot to lift beams, climb stairs, and drive a car."— also posted below.

After three years, two months, and 650,000 lines of code, a team of researchers from MIT’s Computer Science and Artificial Intelligence Lab (CSAIL) stood proudly with their humanoid robot in a sporting arena surrounded by thousands of spectators. They were just one step away from winning the $2 million grand prize at an international competition that many have been calling “the Robot Olympics.”

Granted, that last step was a rather large one: getting a 6’2’’, 400-pound humanoid robot to open a door, rotate a valve, turn on a power tool, drill a hole in a wall, climb stairs, scramble over cinder blocks, and drive a car — all in the space of an hour.

“This is, without a doubt, the most ambitious project that any of us have ever undertaken,” says Professor Russ Tedrake, the CSAIL principal investigator who led the team’s efforts at the DARPA Robotics Challenge (DRC) finals, which took place June 5-6 at the Fairplex in Pomona, California. “From perception to motion-planning to manipulation, the breadth and depth of challenges have forced us to think creatively, program nimbly — and sleep sporadically.”

The CSAIL team finished mere inches away from winning the competition, ultimately earning sixth place out of 25 teams. The Pentagon-sponsored event was aimed at developing mobile robots to perform useful tasks in disaster-relief situations, and was inspired by the 2011 Fukushima nuclear disaster.

For the finals, the researchers had to program their robots to do the same eight tasks as in the December 2013 trials, but without safety tethers, power cords, or consistent communications links.

Regular network blackouts, which were meant to simulate an emergency scenario, greatly benefited the autonomy-minded MIT team: For most tasks, their researchers had planned high-level actions that were partitioned into smaller “chunks” that the robot then executed based on existing templates.

"When you have outages of 30 to 45 seconds, being able to move continuously is vital if you want to avoid losing precious time,” says CSAIL postdoc Scott Kuindersma, who served as MIT’s planning and control lead.

Each team had two chances to finish the tasks as quickly as possible. On MIT’s first run, the team's Boston Dynamics-built Atlas robot broke its right arm trying to get out of the car after the driving task. Despite that setback, the team finished with seven out of the eight points by programming the robot to do all of the remaining tasks left-handed.

On the team’s second run, Atlas sprinted through the first three tasks faster than any other team, and was making extremely strong progress on the drill task until its drill slipped out from the wall, causing another spill and ending the team’s chances to finish first.

“Recovering from the broken arm was a highlight for me,” Kuindersma says. “It was an unplanned demonstration of the adaptiveness of our software and the amazing people behind it."

In the months leading up to the event, researchers logged more than a few 20-hour days, toiling in the tight quarters of a little-used warehouse on Albany Street in Cambridge to hand-code hundreds of complex algorithms for motion planning, perception, and control.

The end product was a comprehensive software system that gathers large swaths of data about the robot’s surroundings to understand its location, create motion plans, and compute how its 28 distinct joints have to move in order to perform the different manipulations.

Researchers developed algorithms for every last one of the robot’s many micro-actions, including taking footsteps, avoiding collisions, performing during network outages, and maintaining balance at such a level that it could stand on one foot while getting out of a five-foot-wide utility vehicle.

In contrast to popular approaches that move a robot incrementally towards a goal, the MIT team opted to adopt a method called “whole-body motion-planning,” which solves for an entire motion in advance.

“When you’re working in an environment that’s filled with uncertainty, programming a robot to slowly nudge towards a goal can mean getting stuck when new obstacles pop up,” says CSAIL PhD student Andrés Valenzuela. “With our approach, we can be extremely confident that we will finish the task that we set out to execute.”

That said, the team experienced more than its fair share of hiccups along the way. The week before the final competition, some of Atlas’ leg joints had started wearing out, forcing researchers to calibrate its motions to be more conservative.

A few months earlier, it fell off a cinderblock and sheared off its entire right arm.

“We tend to lose sight of the fact that, as humans, our systems have been trained to perform these kinds of actions for decades,” Valenzuela says. “It’s little wonder that it’s so difficult to get robots to do these same tasks after three years.”

Where many of their competitors were industry companies with dedicated full-time employees, the MIT team consisted of a professor, a postdoc, and less than a dozen graduate students. “One wrong move would mean the end of our run,” Tedrake says with a raised eyebrow. “Not to mention, a $2 million mistake.” (Fortunately, MIT's two falls didn’t result in any serious damage to the pricey robot.)

Previous DARPA competitions have often spurred significant innovation in both academia and industry: In the years since DARPA’s 2007 Urban Challenge, companies such as Google, Uber, and Apple have begun investing heavily in driverless cars.

Similarly, Tedrake sees the DRC as a tipping point that will inspire more researchers to develop robots that can assist humans with dangerous tasks.

“It’s no longer a question of whether we’ll ever live in a world filled with useful robots,” Tedrake says. “The real question is just how soon that time will come.”

If a widely held assumption about computational complexity is correct, the problem of measuring the difference between two genomes — or texts, or speech samples, or anything else that can be represented as a string of symbols — can’t be solved more efficiently. Piotr Indyk, professor of computer science and engineering.

Read more in the June 11, 2015 MIT News Office article by Larry Hardesty titled "Longstanding problem put to rest - Proof that a 40-year-old algorithm is the best possible will come as a relief to computer scientists." - also posted below.

Comparing the genomes of different species — or different members of the same species — is the basis of a great deal of modern biology. DNA sequences that are conserved across species are likely to be functionally important, while variations between members of the same species can indicate different susceptibilities to disease.

The basic algorithm for determining how much two sequences of symbols have in common — the “edit distance” between them — is now more than 40 years old. And for more than 40 years, computer science researchers have been trying to improve upon it, without much success.

At the ACM Symposium on Theory of Computing (STOC) next week, MIT researchers will report that, in all likelihood, that’s because the algorithm is as good as it gets. If a widely held assumption about computational complexity is correct, then the problem of measuring the difference between two genomes — or texts, or speech samples, or anything else that can be represented as a string of symbols — can’t be solved more efficiently.

In a sense, that’s disappointing, since a computer running the existing algorithm would take 1,000 years to exhaustively compare two human genomes. But it also means that computer scientists can stop agonizing about whether they can do better.

“This edit distance is something that I’ve been trying to get better algorithms for since I was a graduate student, in the mid-’90s,” says Piotr Indyk, a professor of computer science and engineering at MIT and a co-author of the STOC paper. “I certainly spent lots of late nights on that — without any progress whatsoever. So at least now there’s a feeling of closure. The problem can be put to sleep.”

Moreover, Indyk says, even though the paper hasn’t officially been presented yet, it’s already spawned two follow-up papers, which apply its approach to related problems. “There is a technical aspect of this paper, a certain gadget construction, that turns out to be very useful for other purposes as well,” Indyk says.

Squaring off

Edit distance is the minimum number of edits — deletions, insertions, and substitutions — required to turn one string into another. The standard algorithm for determining edit distance, known as the Wagner-Fischer algorithm, assigns each symbol of one string to a column in a giant grid and each symbol of the other string to a row. Then, starting in the upper left-hand corner and flooding diagonally across the grid, it fills in each square with the number of edits required to turn the string ending with the corresponding column into the string ending with the corresponding row.

Computer scientists measure algorithmic efficiency as computation time relative to the number of elements the algorithm manipulates. Since the Wagner-Fischer algorithm has to fill in every square of its grid, its running time is proportional to the product of the lengths of the two strings it’s considering. Double the lengths of the strings, and the running time quadruples. In computer parlance, the algorithm runs in quadratic time.

That may not sound terribly efficient, but quadratic time is much better than exponential time, which means that running time is proportional to 2N, where N is the number of elements the algorithm manipulates. If on some machine a quadratic-time algorithm took, say, a hundredth of a second to process 100 elements, an exponential-time algorithm would take about 100 quintillion years.

Theoretical computer science is particularly concerned with a class of problems known as NP-complete. Most researchers believe that NP-complete problems take exponential time to solve, but no one’s been able to prove it. In their STOC paper, Indyk and his student Artūrs Bačkurs demonstrate that if it’s possible to solve the edit-distance problem in less-than-quadratic time, then it’s possible to solve an NP-complete problem in less-than-exponential time. Most researchers in the computational-complexity community will take that as strong evidence that no subquadratic solution to the edit-distance problem exists.

Can’t get no satisfaction

The core NP-complete problem is known as the “satisfiability problem”: Given a host of logical constraints, is it possible to satisfy them all? For instance, say you’re throwing a dinner party, and you’re trying to decide whom to invite. You may face a number of constraints: Either Alice or Bob will have to stay home with the kids, so they can’t both come; if you invite Cindy and Dave, you’ll have to invite the rest of the book club, or they’ll know they were excluded; Ellen will bring either her husband, Fred, or her lover, George, but not both; and so on. Is there an invitation list that meets all those constraints?

In Indyk and Bačkurs’ proof, they propose that, faced with a satisfiability problem, you split the variables into two groups of roughly equivalent size: Alice, Bob, and Cindy go into one, but Walt, Yvonne, and Zack go into the other. Then, for each group, you solve for all the pertinent constraints. This could be a massively complex calculation, but not nearly as complex as solving for the group as a whole. If, for instance, Alice has a restraining order out on Zack, it doesn’t matter, because they fall in separate subgroups: It’s a constraint that doesn’t have to be met.

At this point, the problem of reconciling the solutions for the two subgroups — factoring in constraints like Alice’s restraining order — becomes a version of the edit-distance problem. And if it were possible to solve the edit-distance problem in subquadratic time, it would be possible to solve the satisfiability problem in subexponential time.

“This is really nice work,” says Barna Saha, an assistant professor of computer science at the University of Massachusetts atAmherst. “There are lots of people who have been working on this problem, because it has a big practical impact. But they won’t keep trying to develop a subquadratic algorithm, because that seems very unlikely to happen, given the result of this paper.”

As for the conjecture that the MIT researchers’ proof depends on — that NP-complete problems can’t be solved in subexponential time — “It’s a very widely believed conjecture,” Saha says. “And there are many other results in this low-polynomial-time complexity domain that rely on this conjecture.

DARPA Robotics Challenge MIT Team leader Russ Tedrake reports on the real win in the team's sixth placement in last week's competition. The team not only won the overall best-paper award at the 2014 International Conference on Humanoid Robots, but they also accomplished research that will have huge near and longterm payoffs.

Read more in the June 11, 2015 MIT News Office article by Larry Hardesty titled "Robotics competition generated groundbreaking research - MIT team based robotic control algorithms on cutting-edge theory." - also posted below. [Image: A still from a video of MIT's first DARPA Robotics Challenge run, where they scored 7 points.] Courtesy of the MITDRC team

Last weekend was the final round of competition in the U.S. Defense Advanced Research Projects Agency’s contest to design control systems for a humanoid robot that could climb a ladder, remove debris, drive a utility vehicle, and perform several other tasks related to a hypothetical disaster. The team representing MIT finished sixth out of a field of 25.

But before the competition, the team’s leader, Russ Tedrake, an associate professor of computer science and engineering, said, “I feel as if we’ve already won, because of all the amazing research our students did” — including a paper that won the overall best-paper award at the 2014 International Conference on Humanoid Robots.

Optima primed

In control theory, control of a dynamic system — such as a robot, an airplane, or a power grid — is often treated as an optimization problem. The trick is to contrive a mathematical function whose minimum value represents a desired state of the system. Control is then a matter of finding that minimum and figuring out how to continuously nudge the system back toward it.

Optimization problems can be enormously complex, so they’re frequently used for offline analysis — for example, to determine how well much simpler control algorithms will work. But from the get-go, Tedrake decided that the MIT team’s control algorithms would solve optimization problems on the fly. That required innovation on multiple fronts. [Video: See how the MIT team designed their robot to compete in the DARPA Robotics Challenge. Video: CSAIL.]

Balancing act

The lower-level control algorithm, however, can’t afford to ignore the forces acting at individual points of contact. Early on, Tedrake set the ambitious goal of a system that could evaluate information from the robot’s sensors and readjust the trajectories of its limbs 1,000 times a second, or at a rate of one kilohertz.

That sounds daunting, but as Tedrake explains, past a certain point, the high sampling rate actually becomes an advantage. One one-thousandth of a second allows so little time for circumstances to change that the imposition of new constraints usually occurs piecemeal. From one sensor reading to the next, the algorithm rarely has to meet more than one or two new constraints, which it can usually manage with just a small adjustment.

As one test of the kilohertz controller, members of the MIT team instructed their robot to dismount from the utility vehicle they’d been using to test its driving skills; once it had transferred all its weight to one foot, they started jumping up and down on the vehicle’s fenders. The robot maintained its balance.

Human factors

For several of the robot’s tasks, the MIT researchers exploited the fact that the contest allowed human operators to communicate with their robots — although their communication links would be erratic.

Although the robot has an onboard camera, its chief sensor is a laser rangefinder, which fires pulses of light in different directions and measures the time they take to return. This produces a huge cloud of individual points — some of which belong to the same objects, and some of which don’t. Resolving that point cloud into distinct objects is an extremely difficult task, which computer vision researchers have been wrestling with for decades. It would be almost impossible to perform in real time.

So the MIT researchers built a library of generic geometric representations of objects the robot was likely to encounter — such as the fallen lumber whose removal was one of its tasks during the competition finals. The remote operator can look at an image captured by the robot’s camera, identify the appropriate library of object representations, and superimpose the point cloud produced by the laser rangefinder. Then the operator clicks the track pad or mouse button twice to roughly indicate the ends of the objects in the image. Algorithms then automatically cluster points together according to the geometric models, picking out the individual objects that the robot will have to manipulate.

When the robot enters a new environment, its rangefinder readings can tell it where nearby objects are. But it doesn’t know which are safe to step on. So the MIT researchers also developed an interface that allows the robot’s operator to click on a graphical representation of the robot’s surroundings, identifying flat surfaces that offer secure footholds.

From the robot’s sensor readings, the algorithm automatically determines the extent of the safe areas, by locating the first significant changes in altitude. So if the operator clicks at a single point on an uncluttered floor, the interface highlights an expanse of space that extends outward from that point to the first obstacles the rangefinder registers. Similarly, if the operator clicks a single point on one step of a staircase, the algorithm highlights most of the rest of the step, but stops short of its edges.

“If you look at what happened at DRC [DARPA Robotics Challenge], it was a lot of teleoperation, a lot of scripted pieces of movement, and then a human telling the robot which movement to execute in great detail,” says Emanuel Todorov, an associate professor of electrical engineering and computer science at the University of Washington. “Humans are smart, and at least for the time being, if you put them in the loop, they outperform the autonomous controllers Russ and others built. But eventually it’s going to turn the other way around, because these are complicated machines, and there’s only so much a human can figure out in real time. The approach that Russ was taking was in some sense the right approach. This is what robotics should look like five or 10 year from now.”

At the recent International Conference on Robotics and Automation, MIT researchers led by Daniela Rus, the Andrew and Erna Viterbi Professor in MIT’s Department of Electrical Engineering and Computer Science,  presented a printable origami robot that folds itself up from a flat sheet of plastic when heated and measures about a centimeter from front to back.

Read more in the June 12, 2015 MIT News Office article by Larry Hardesty titled "Centimeter-long origami robot - Controlled by magnetic fields, tiny robot climbs inclines, swims, and carries loads twice its weight." - also posted below.

Weighing only a third of a gram, the robot can swim, climb an incline, traverse rough terrain, and carry a load twice its weight. Other than the self-folding plastic sheet, the robot’s only component is a permanent magnet affixed to its back. Its motions are controlled by external magnetic fields.

"The entire walking motion is embedded into the mechanics of the robot body,” says Cynthia R. Sung, an MIT graduate student in electrical engineering and computer science and one of the robot’s co-developers. “In previous [origami] robots, they had to design electronics and motors to actuate the body itself."

Joining Sung on the paper describing the robot are her advisor, Daniela Rus, the Andrew and Erna Viterbi Professor in MIT’s Department of Electrical Engineering and Computer Science; first author Shuhei Miyashita, a postdoc in Rus’ lab; Steven Guitron, who just received his bachelor’s degree in mechanical engineering from MIT; and Marvin Ludersdorfer of the Technical University of Munich.

Fantastic Voyage

The robot’s design was motivated by a hypothetical application in which tiny sheets of material would be injected into the human body, navigate to an intervention site, fold themselves up, and, when they had finished their assigned tasks, dissolve. To that end, the researchers built their prototypes from liquid-soluble materials. One prototype robot dissolved almost entirely in acetone (the permanent magnet remained); another had components that were soluble in water.

“We complete the cycle from birth through life, activity, and the end of life,” Miyashita says. “The circle is closed.”

In all of the researchers’ prototypes, the self-folding sheets had three layers. The middle layer always consisted of polyvinyl chloride, a plastic commonly used in plumbing pipes, which contracts when heated. In the acetone-soluble prototype, the outer layers were polystyrene.

Slits cut into the outer layers by a laser cutter guide the folding process. If two slits on opposite sides of the sheet are of different widths, then when the middle layer contracts, it forces the narrower slit’s edges together, and the sheet bends in the opposite direction. In their experiments, the researchers found that the sheet would begin folding at about 150 degrees Fahrenheit.

Once the robot has folded itself up, the proper application of a magnetic field to the permanent magnet on its back causes its body to flex. The friction between the robot’s front feet and the ground is great enough that the front feet stay fixed while the back feet lift. Then, another sequence of magnetic fields causes the robot’s body to twist slightly, which breaks the front feet’s adhesion, and the robot moves forward.

Outside control

In their experiments, the researchers positioned the robot on a rectangular stage with an electromagnet at each of its four corners. They were able to vary the strength of the electromagnets’ fields rapidly enough that the robot could move nearly four body lengths a second.

In addition to the liquid-soluble versions of their robot, the researchers also built a prototype whose outer layers were electrically conductive. Inspired by earlier work from Rus and Miyashita, the researchers envision that a tiny, conductive robot could act as a sensor. Contact with other objects — whether chemical accretions in a mechanical system or microorganisms or cells in the body — would disrupt a current passing through the robot in a characteristic way, and that electrical signal could be relayed to human operators.

“Making small robots is particularly challenging, because you don’t just take off-the-shelf components and bolt them together,” says Hod Lipson, a professor of mechanical and aerospace engineering at Cornell University, who studies robotics. “It’s a challenging angle of robotics, and they’ve been able to solve it.”

Press coverage:

betaBoston, Nidhi Subbaraman

Newsweek, Lauren Walker

CBC News, Lauren O'Neill

EECS Department Head Anantha Chandrakasan announced today the appointment of Christopher J. Terman as the new EECS Undergraduate Officer. Terman succeeds Albert Meyer, the Hitachi America Professor of Engineering, who is ending his term in the position. Terman previously served as Undergraduate Officer from 2009–2011.

Terman has been a member of the EECS for thirty-two of the last forty-two years, first as a graduate student, then as a member of the faculty. Returning to MIT after 10 years as an entrepreneur, he became a Senior Lecturer in 1997, subsequently pursuing his passion for teaching, mostly focused on the architecture and implementation of digital systems (6.02, 6.004, 6.099, 6.111, 6.173, 6.371). He has won a number of teaching awards, including the 2008 Jamieson Prize for Excellence in Teaching. Terman previously served as Undergraduate Officer (2009–2011) and he played an active role in the administration of CSAIL as both Associate Director and Co-Director. He was the Research Director for the T-Party and Qmulus projects, a ten-year, $45 million research collaboration with Quanta Computer. His research interests include computer-aided design tools and educational technologies.

"I would like to take this opportunity to express my deepest thanks to Professor Albert Meyer for his outstanding contributions as the EECS Undergraduate Officer over the past two years," Chandrakasan wrote in an email announcing the transition to EECS Faculty. "I look forward to working with Chris in his new capacity and with Albert as he continues to assist the department in several strategic areas." 

Albert Meyer oversaw the EECS Undergraduate Office from 2013 to 2015 through two years of successively record-breaking enrollment increases. Additionally, he enlisted the HKN and IEEE student organizations to support the advising system by offering advising presentations by students and staff on topics such as beginning-of-term orientations, drop-date decisions, and choice of AUS. As Undergraduate Officer, Meyer directed the development of advisor, student, and undergraduate administrative portals. The portals provide convenient web access to up-to-date department audits for individual students and advisors, as well as student and advising information from multiple sources. In particular, the portal has significantly reduced Undergraduate Office time to clear students for the degree list. In addition, he designed and oversaw the installation of improved MEng application and thesis submission processes, ensuring that MEng theses are deposited in DSpace. Meyer also designed and oversaw refinements of the department audit form and grades meeting display system. He recruited and trained a new CS Transfer Credit approver while he continued to serve as CS Substitution officer, as well as ex-officio on the ABET, Curriculum, and USAGE Committees.

Larry Hardesty | MIT News Office

Video-processing algorithm magnifies motions indiscernible to naked eye, even in moving objects.

(Image: A key to the new motion-magnification algorithm is a technique for very precisely extracting foreground objects from their backgrounds. Courtesy of the researchers.)

For several years now, the research groups of MIT professors of computer science and engineering William Freeman and Frédo Durand have been investigating techniques for amplifying movements captured by video but indiscernible to the human eye. Versions of their algorithms can make the human pulse visible and even recover intelligible speech from the vibrations of objects filmed through soundproof glass.

Earlier this month, at the Computer Vision and Pattern Recognition conference, Freeman, Durand, and colleagues at the Qatar Computing Research Institute (QCRI) presented a new version of the algorithm that can amplify small motions even when they’re contained within objects executing large motions. So, for instance, it could make visible the precise sequence of muscle contractions in the arms of a baseball player swinging the bat, or in the legs of a soccer player taking a corner kick.

“The previous version of the algorithm assumed everything was small in the video,” Durand says. “Now we want to be able magnify small motions that are hidden within large motions. The basic idea is to try to cancel the large motion and go back to the previous situation.”

Canceling the large motion means determining which pixels of successive frames of video belong to a moving object and which belong to the background. As Durand explains, that problem becomes particularly acute at the object’s boundaries.

If a digital camera captures an image of, say, a red object against a blue background, some of its photosensors will register red light, and some will register blue. But the sensors corresponding to the object’s boundaries may in fact receive light from both foreground and background, so they’ll register varying shades of purple.

Ordinarily, an algorithm separating foreground from background could probably get away with keeping those borderline pixels: A human viewer probably wouldn’t notice a tiny fringe of purple around a red object. But the purpose of the MIT researchers’ motion amplification algorithm is precisely to detect variations invisible to the naked eye. Changes of color at an object’s boundaries could be interpreted as motions requiring magnification.

So Durand, Freeman, and Mohamed Elgharib and Mohamed Hefeeda of QCRI instead assign each boundary pixel a weight, corresponding to the likelihood that it belongs to the foreground object. In the example of the red object against a blue background, that weight would simply depend on whether the shade of purple is bluer or redder. Then, on the basis of the pixels’ weights, the algorithm randomly discards some and keeps others. On average, it will make the right decision, and it will disrupt any patterns of color change that could be mistaken for motion.

The problem of identifying the same object from frame to frame, Durand says, is related to the problem of image stabilization, which attempts to remove camera jitter from video. Identifying the motion of a single object, however, is more difficult than determining the motion of the image as a whole.

The MIT and QCRI researchers make a few assumptions to render the problem more tractable. First, they assume a correlation between the direction and rate of motion of adjacent pixels. Second, they assume “smoothness” — that the direction and rate of motion will be consistent over time. Finally, they assume that pixels’ trajectories across frames can be captured by linear mathematical relationships, which enables their algorithm to analyze pixels individually.

Then, rather than looking for correlations between one frame and the next, their algorithm considers five frames at a time, using consistencies across frames to resolve ambiguities between adjacent frames.

Once the algorithm has identified the pixels correlating to a single moving object, it corrects for the object’s motion and performs the same motion magnification procedure that previous versions did. Finally, it reinserts the magnified motions back into the original video stream.

Learn more about the Computer Graphics Group

Read this article on MIT News

Larry Hardesty | MIT News Office

New ultralow-power circuit improves efficiency of energy harvesting to more than 80 percent.

(Image: The MIT researchers' prototype for a chip measuring 3 millimeters by 3 millimeters. The magnified detail shows the chip's main control circuitry, including the startup electronics; the controller that determines whether to charge the battery, power a device, or both; and the array of switches that control current flow to an external inductor coil. This active area measures just 2.2 millimeters by 1.1 millimeters. Courtesy of the researchers.)

The latest buzz in the information technology industry regards “the Internet of things” — the idea that vehicles, appliances, civil-engineering structures, manufacturing equipment, and even livestock would have their own embedded sensors that report information directly to networked servers, aiding with maintenance and the coordination of tasks.

Realizing that vision, however, will require extremely low-power sensors that can run for months without battery changes — or, even better, that can extract energy from the environment to recharge.

Last week, at the Symposia on VLSI Technology and Circuits, MIT researchers presented a new power converter chip that can harvest more than 80 percent of the energy trickling into it, even at the extremely low power levels characteristic of tiny solar cells. Previous ultralow-power converters that used the same approach had efficiencies of only 40 or 50 percent.

Moreover, the researchers’ chip achieves those efficiency improvements while assuming additional responsibilities. Where most of its ultralow-power predecessors could use a solar cell to either charge a battery or directly power a device, this new chip can do both, and it can power the device directly from the battery.

All of those operations also share a single inductor — the chip’s main electrical component — which saves on circuit board space but increases the circuit complexity even further. Nonetheless, the chip’s power consumption remains low. “We still want to have battery-charging capability, and we still want to provide a regulated output voltage,” says Dina Reda El-Damak, an MIT graduate student in electrical engineering and computer science and first author on the new paper. “We need to regulate the input to extract the maximum power, and we really want to do all these tasks with inductor sharing and see which operational mode is the best. And we want to do it without compromising the performance, at very limited input power levels — 10 nanowatts to 1 microwatt — for the Internet of things.” The prototype chip was manufactured through the Taiwan Semiconductor Manufacturing Company's University Shuttle Program.

Ups and downs

The circuit’s chief function is to regulate the voltages between the solar cell, the battery, and the device the cell is powering. If the battery operates for too long at a voltage that’s either too high or too low, for instance, its chemical reactants break down, and it loses the ability to hold a charge. To control the current flow across their chip, El-Damak and her advisor, Anantha Chandrakasan, the Joseph F. and Nancy P. Keithley Professor in Electrical Engineering, use an inductor, which is a wire wound into a coil. When a current passes through an inductor, it generates a magnetic field, which in turn resists any change in the current.

Throwing switches in the inductor’s path causes it to alternately charge and discharge, so that the current flowing through it continuously ramps up and then drops back down to zero. Keeping a lid on the current improves the circuit’s efficiency, since the rate at which it dissipates energy as heat is proportional to the square of the current.

Once the current drops to zero, however, the switches in the inductor’s path need to be thrown immediately; otherwise, current could begin to flow through the circuit in the wrong direction, which would drastically diminish its efficiency. The complication is that the rate at which the current rises and falls depends on the voltage generated by the solar cell, which is highly variable. So the timing of the switch throws has to vary, too.

Electric hourglass

To control the switches’ timing, El-Damak and Chandrakasan use an electrical component called a capacitor, which can store electrical charge. The higher the current, the more rapidly the capacitor fills. When it’s full, the circuit stops charging the inductor.

The rate at which the current drops off, however, depends on the output voltage, whose regulation is the very purpose of the chip. Since that voltage is fixed, the variation in timing has to come from variation in capacitance. El-Damak and Chandrakasan thus equip their chip with a bank of capacitors of different sizes. As the current drops, it charges a subset of those capacitors, whose selection is determined by the solar cell’s voltage. Once again, when the capacitor fills, the switches in the inductor’s path are flipped.

“In this technology space, there’s usually a trend to lower efficiency as the power gets lower, because there’s a fixed amount of energy that’s consumed by doing the work,” says Brett Miwa, who leads a power conversion development project as a fellow at the chip manufacturer Maxim Integrated. “If you’re only coming in with a small amount, it’s hard to get most of it out, because you lose more as a percentage. [El-Damak’s] design is unusually efficient for how low a power level she’s at.”

“One of the things that’s most notable about it is that it’s really a fairly complete system,” he adds. “It’s really kind of a full system-on-a chip for power management. And that makes it a little more complicated, a little bit larger, and a little bit more comprehensive than some of the other designs that might be reported in the literature. So for her to still achieve these high-performance specs in a much more sophisticated system is also noteworthy.”

Learn more about the Chandrakasan Group

Read this article on MIT News

A high-school outreach program first brought Tamara Broderick to MIT in 2002. Now she's back, as an assistant professor in EECS.

(Image: Tamara Broderick. Image credit: Lillie Paquette/School of Engineering)

Tamara Broderick’s career at MIT started when she was still in high school. In 2002, she participated in the inaugural class of MIT’s Women's Technology Program (WTP)— a new experiment at the time that brought high school girls to campus for a four-week program designed to spark a passion for engineering and computer science. She arrived on campus that summer from Cleveland, Ohio, thrilled at the prospect of an immersion in science and math. “WTP was an amazing opportunity to learn the basics of computer science, electrical engineering, and applied math with like-minded peers,” she says. “I felt I was suddenly in a place where I really belonged.”

Now the ITT Career Development Assistant Professor in the Department of Electrical Engineering and Computer Science, Broderick recalls “open-ended projects that pushed frontiers for me,” such as biopsying a grape suspended in Jell-O with a homemade Lego-based probe. There was extracurricular fun as well with a “group of people who shared the same nerdy sensibility.” She saw the Boston Pops, ate her first dim sum, and drew the digits of pi on a beach. But most of all Broderick credits WTP with demonstrating the potential of computer science and mathematics as fields of study. “My experience at WTP was formative,” she says. “It had a broad impact on what I wanted to do, in my decision to pursue math and computer science more generally.”

Broderick set out to learn more about physics, math, and computer science, earning a BA in mathematics from Princeton University in 2007. She went on to earn a master of advanced study in mathematics in 2008 and an MPhil in physics in 2009 at Cambridge University. While on a Marshall Scholarship in the UK, Broderick discovered Bayesian statistics. “A lot of my interests coalesced,” she says. She then came back to the U.S. in 2009, to the University of California at Berkeley, to begin a master’s in computer science (which she completed in 2013) and a PhD in statistics, which she earned in 2014.

Her doctoral work in machine learning involved developing algorithms that can speed computational analysis of large, streaming data sets, enabling computers to discover more and more hidden and meaningful structure in data as more data is obtained — without the need for prior human annotation.

Though Broderick describes her work as primarily theoretical, it is already lending itself to useful and surprising applications. One of her algorithms has been deployed to analyze tumor heterogeneity. Different types of cancer may appear in a single tumor, and identifying the disparate yet connected types of diseased tissue may speed a more targeted treatment approach.

This algorithm and the theory behind it was the work Broderick presented at MIT in 2013 during another recruitment program devoted to increasing gender diversity, Rising Stars. Initially developed in the Department of Aeronautics and Astronautics, the program showcases the research of female PhDs and postdoctoral candidates from around the U.S. Rising Stars gave Broderick a second taste of the dynamic academic environment that had made such an impression on her a decade earlier. “There are so many amazing people at MIT at the top of their game,” she says. “And it’s a place that helps you start exchanging ideas, and shows you that research isn’t a thing you do alone, but with a lot of other people.”

Broderick’s focus on machine learning and statistics, on display at this workshop, opened the door to her current position. On top of assignments that include a new graduate course in theoretical statistics, and work for MIT’s new Institute for Data, Systems, and Society, Broderick has an additional item on her agenda: “I’m giving a talk at WTP this summer,” she says, “introducing students to the kind of research I do.”

Read this article on MIT News