ActiSonic.com is for Sale

Unraveling the Matrix

Among the most common tools in electrical engineering and computer science are rectangular grids of numbers known as matrices. The numbers in a matrix can represent data: The rows, for instance, could represent temperature, air pressure and humidity, and the columns could represent different locations where those three measurements were taken. But matrices can also represent mathematical equations. If the expressions t + 2p + 3h and 4t + 5p + 6h described two different mathematical operations involving temperature, pressure and humidity measurements, they could be represented as a matrix with two rows, [1 2 3] and [4 5 6]. Multiplying the two matrices together means performing both mathematical operations on every column of the data matrix and entering the results in a new matrix. In many time-sensitive engineering applications, multiplying matrices can give quick but good approximations of much more complicated calculations.

In a paper published in the July 13 issue of Proceedings of the National Academy of Science, MIT math professor Gilbert Strang describes a new way to split certain types of matrices into simpler matrices. The result could have implications for software that processes video or audio data, for compression software that squeezes down digital files so that they take up less space, or even for systems that control mechanical devices.

Strang?s analysis applies to so-called banded matrices. Most of the numbers in a banded matrix are zeroes; the only exceptions fall along diagonal bands, at or near the central diagonal of the matrix. This may sound like an esoteric property, but it often has practical implications. Some applications that process video or audio signals, for instance, use banded matrices in which each band represents a different time slice of the signal. By analyzing local properties of the signal, the application could, for instance, sharpen frames of video, or look for redundant information that can be removed to save memory or bandwidth.

Working backwards

Since most of the entries in a banded matrix ? maybe 99 percent, Strang says ? are zero, multiplying it by another matrix is a very efficient procedure: You can ignore all the zero entries. After a signal has been processed, however, it has to be converted back into its original form. That requires multiplying it by the ?inverse? of the processing matrix: If multiplying matrix A by matrix B yields matrix C, multiplying C by the inverse of B yields A.

But the fact that a matrix is banded doesn?t mean that its inverse is. In fact, Strang says, the inverse of a banded matrix is almost always ?full,? meaning that almost all of its entries are nonzero. In a signal-processing application, all the speed advantages offered by banded matrices would be lost if restoring the signal required multiplying it by a full matrix. So engineers are interested in banded matrices with banded inverses, but which matrices those are is by no means obvious.

In his PNAS paper, Strang describes a new technique for breaking a banded matrix up into simpler matrices ? matrices with fewer bands. It?s easy to tell whether these simpler matrices have banded inverses, and if they do, their combination will, too. Strang?s technique thus allows engineers to determine whether some promising new signal-processing techniques will, in fact, be practical.

Faster than Fourier?

One of the most common digital-signal-processing techniques is the discrete Fourier transform (DFT), which breaks a signal into its component frequencies and can be represented as a matrix. Although the matrix for the Fourier transform is full, Strang says, ?the great fact about the Fourier transform is that it happens to be possible, even though it?s full, to multiply fast and to invert it fast. That?s part of what makes Fourier wonderful.? Nonetheless, for some signal-processing applications, banded matrices could prove more efficient than the Fourier transform. If only parts of the signal are interesting, the bands provide a way to home in on them and ignore the rest. ?Fourier transform looks at the whole signal at once,? Strang says. ?And that?s not always great, because often the signal is boring for 99 percent of the time.?

Richard Brualdi, the emeritus UWF Beckwith Bascom Professor of Mathematics at the University of Wisconsin-Madison, points out that a mathematical conjecture that Strang presents in the paper has already been proven by three other groups of researchers. ?It?s a very interesting theorem,? says Brualdi. ?It?s already generated a couple of papers, and it?ll probably generate some more.? Brualdi points out that large data sets, such as those generated by gene sequencing, medical imaging, or weather monitoring, often yield matrices with regular structures. Bandedness is one type of structure, but there are others, and Brualdi expects other mathematicians to apply techniques like Strang?s to other types of structured matrices. ?Whether or not those things will work, I really don?t know,? Brualdi says. ?But Gil?s already said that he?s going to look at a different structure in a future paper.?

Super-sizing a cancer drug minimizes side effects

One of the first chemotherapy drugs given to patients diagnosed with cancer ? especially lung, ovarian or breast cancer ? is cisplatin, a platinum-containing compound that gums up tumor cells? DNA. Cisplatin does a good job of killing those tumor cells, but it can also seriously damage the kidneys, which receive high doses of cisplatin because they filter the blood.

Now a team of scientists at the Harvard-MIT Division of Health Sciences and Technology (HST) has come up with a new way to package cisplatin into nanoparticles that are too big to enter the kidneys. The new compound could spare patients the usual side effects and allow doctors to administer higher doses of the drug, says Shiladitya Sengupta, leader of the research team.

?We could give so much more cisplatin than is now possible,? says Sengupta, an assistant professor of HST. ?You could wipe out the tumor by carpet-bombing it.?

Tumors in mice treated with the new cisplatin nanoparticle shrank to half the size of those treated with traditional cisplatin, with minimal side effects. The findings were reported in the Proceedings of the National Academy of Sciences in June.

Beads on a string

Doctors began using cisplatin to treat cancer in the 1970s. Early on, doctors recognized that it harmed the kidneys, and cancer researchers began looking for alternatives. In the past few decades, the FDA has approved two less-toxic derivatives of cisplatin: carboplatin and oxaliplatin. However, those drugs don?t kill tumor cells as successfully as cisplatin.

Cisplatin?s effectiveness lies in how easily it releases its platinum molecule, freeing it to cross-link DNA strands, disrupting cell division and forcing the cell to undergo suicide. Carboplatin and oxaliplatin are less effective (but less toxic) than cisplatin because they hold on to their platinum atoms more tightly.

Sengupta and his colleagues took a new approach to making cisplatin safer: stringing cisplatin molecules together into a nanoparticle that is too large to get into the kidneys. (It has been shown that the kidneys cannot absorb particles larger than five nanometers ? about 1/10,000th the diameter of a human hair).

His team designed a polymer that binds to cisplatin, arranging the molecules like beads on a string. The string then winds itself into a nanoparticle about 100 nanometers long ? much too large to fit into the kidneys. However, the particles can still reach tumor cells because tumors are surrounded by ?leaky? blood vessels, which have 500-nanometer pores.

Their first nanoparticle proved less effective than cisplatin, so they tweaked the polymer to make it hold a little less tightly to platinum, and ended up with a molecule with a tumor-killing power similar to cisplatin?s. However, because its side effects are minimal, the nanoparticle can be delivered in higher doses.

Daniela Dinulescu, an author of the paper and pathology instructor at Brigham and Women?s Hospital in Boston, showed that the nanoparticles outperformed cisplatin in mice engineered to develop ovarian cancer. The researchers also showed it to be effective against lung and breast tumor cells grown in the lab. Once the tumor cells die, the immune system clears platinum from the body.

The research was funded by the Department of Defense Breast Cancer Research Program and the National Institutes of Health.

It is difficult to develop and gain approval for new platinum-based compounds, says Nicholas Farrell, professor of inorganic chemistry at Virginia Commonwealth University, but he believes Sengupta?s new nanoparticles are promising. ?If successful, the approach promises to maintain the status of cisplatin as one of the most useful drugs available to the clinician,? says Farrell.

The MIT researchers are now working on new variants of the nanoparticles that would be easier to manufacture. They are also making plans to test the nanoparticles in clinical trials, which Sengupta hopes will get underway within the next two years. The polymer used for the nanoparticle backbone is similar to malic acid, a natural product of cellular metabolism, so Sengupta is optimistic that it will prove safe in humans.

Proteins linked to longevity also linked to Alzheimer?s

Over the past 20 years, scientists have learned that proteins called sirtuins play a vital role in longevity and stress response in organisms as diverse as humans, yeast and mice. A new paper from MIT biologists now reveals a surprising additional role for sirtuins: They appear to suppress the production of amyloid beta proteins, which form plaques in the brains of Alzheimer?s patients.

The finding, reported in the July 23 issue of Cell, suggests that targeting sirtuins could offer a promising new approach to treating Alzheimer?s, says Professor Leonard Guarente, leader of the research team.

Guarente and his colleagues showed that boosting the activity of a sirtuin called SIRT1 stifled the production of amyloid beta proteins and enhanced brain function in mice engineered to express Alzheimer?s symptoms. This marks the first time sirtuins have been linked to those proteins.

Several drug companies are now developing and testing compounds that enhance sirtuin activity. Guarente, who consults for one of those companies, Sirtris (a unit of GlaxoSmithKline), believes that sirtuin activators may eventually prove useful against Alzheimer?s, which affects up to one-third of people who reach age 80.

Protein clumping

Though amyloid plaques are a defining feature of Alzheimer?s disease, many researchers now believe that the symptoms are caused by smaller clumps of two or three amyloid beta (A-beta) fragments, not the larger plaques.

A-beta peptides form when proteins called amyloid precursor proteins (APPs) are broken into smaller pieces. However, APPs can also be cleaved at other sites, producing harmless protein fragments. APP?s normal function is unknown, but it has been established that people with a gene mutation that stimulates overproduction of APP are more likely to develop Alzheimer?s at an early age (before age 65).

Another mutation that stimulates early-onset Alzheimer?s (which accounts for 5 to 10 percent of cases) occurs in the gene for the enzyme that cleaves APP into A-beta peptides. Although those genes for early-onset Alzheimer?s have been identified, ?with late-onset Alzheimer?s, we still don?t know why some people get it and other people don?t,? says Guarente.

Guarente, who first discovered the life-extending ability of sirtuins 20 years ago, started studying their role in Alzheimer?s after some recent studies showed that the gene that produces sirtuins, SIRT1, appears to protect mice from the effects of Alzheimer?s disease. When those studies came out, ?I thought that the mice with extra SIRT1 probably had just as much A-beta, but that SIRT1 was protecting them against it,? Guarente recalls. ?It turns out that they were actually making less A-beta peptide.?

In the Cell paper, Guarente and his colleagues showed that SIRT1 activates the production of an enzyme (alpha-secretase) that carves APPs into harmless fragments, preventing the formation of Alzheimer?s-associated amyloid peptides. Mice engineered to produce excess sirtuins had reduced peptide levels, while mice with SIRT1 knocked out showed elevated peptide levels.

Furthermore, learning and memory deficits in the Alzheimer?s mice were improved when SIRT1 was overproduced and worsened when the gene was deleted. The researchers also found that SIRT1 activates the so-called notch-signaling pathway via the elevated levels of alpha-secretase, which protects neurons and helps maintain brain function.

A new target for Alzheimer?s

The research, funded by the American Parkinson Disease Association, National Institutes of Health and the Paul F. Glenn Foundation, demonstrates that drugs that activate SIRT1 in the brain may be a promising approach to treating Alzheimer?s, says Guarente. Any such drug would have to be able to cross the blood-brain barrier, which prevents large molecules from diffusing into the brain.

Sirtris, a company co-founded by Guarente and then bought by GlaxoSmithKline, is now testing SIRT1 activators in a clinical trial for diabetes. Guarente believes that related drugs could have an impact on a range of neurodegenerative diseases, as well as diabetes and other diseases of aging.

However, any potential drug for Alzheimer?s would likely take several years to reach clinical trials, because of the need to find a drug that crosses the blood-brain barrier, says Guarente.

Rudolph Tanzi, professor of neurology at Harvard Medical School, says the new findings also suggest another approach: targeting one specific aspect of SIRT1?s activity. Tanzi?s lab recently found that mutations in the gene that produces alpha-secretase (ADAM10) are associated with late-onset Alzheimer?s disease.

?If this is how SIRT1 protects against Alzheimer?s ? by turning on ADAM10 ? you could try finding a drug that specifically addresses that mechanism,? instead of globally activating SIRT1, says Tanzi. 

RNA offers a safer way to reprogram cells

In recent years, scientists have shown that they can reprogram human skin cells to an immature state that allows the cells to become any type of cell. This ability, known as pluripotency, holds the promise of treating diseases such as diabetes and Parkinson?s disease by transforming the patients? own cells into replacements for the nonfunctioning tissue.

However, the techniques now used to transform cells pose some serious safety hazards. To deliver the genes necessary to reprogram cells to a pluripotent state, scientists use viruses carrying DNA, which then becomes integrated into the cell?s own DNA. But this so-called DNA-based reprogramming carries the risk of disrupting the cell?s genome and leading it to become cancerous.

Now, for the first time, MIT researchers have shown that they can deliver those same reprogramming genes using RNA, the genetic material that normally ferries instructions from DNA to the cell?s protein-making machinery. This method could prove much safer than DNA-based reprogramming, say the researchers, Associate Professor of Electrical and Biological Engineering Mehmet Fatih Yanik and electrical engineering graduate student Matthew Angel.

Yanik and Angel describe the method, also the subject of Angel?s master?s thesis, in the July 23 issue of the journal PLoS ONE.

However, the researchers say they cannot yet claim to have reprogrammed the cells into a pluripotent state. To prove that, they would need to grow the cells in the lab for a longer period of time and study their ability to develop into other cell types ? a process now underway in their lab. Their key achievement is demonstrating that the genes necessary for reprogramming can be delivered with RNA.

?Before this, nobody had a way to transfect cells multiple times with protein-encoding RNA,? says Yanik. (Transfection is the process of introducing DNA or RNA into a cell without using viruses to deliver them.)

Focus on RNA

In 2006, researchers at Kyoto University showed they could reprogram mouse skin cells into a pluripotent, embryonic-like state with just four genes. More recently, other scientists have achieved the same result in human cells by delivering the proteins encoded by those genes directly into mature cells, but that process is more expensive, inefficient and time-consuming than reprogramming with DNA.

Funded by a Packard Fellowship in Science and Engineering, Yanik and Angel decided to pursue a new alternative by transfecting cells with messenger RNA (mRNA), a short-lived molecule that carries genetic instructions copied from DNA.

However, they found that RNA transfection poses a significant challenge: When added to mature human skin cells, mRNA provokes an immune response meant to defend against viruses made of RNA. Repeated exposure to long strands of RNA leads cells to undergo cell suicide, sacrificing themselves to help prevent the rest of the body from being infected.

Yanik and Angel knew that some RNA viruses, including hepatitis C, can successfully suppress that defensive response. After reviewing studies of hepatitis C?s evasive mechanisms, they did experiments showing they could shut off the response by delivering short interfering RNA (siRNA) that blocks production of several proteins key to the response.

Once the defense mechanism is shut off, mRNA carrying the genes for cell reprogramming can be safely delivered. The researchers showed that they could induce cells to produce the reprogramming proteins for more than a week, by delivering siRNA and mRNA every other day.

Peter Andrews, director of the Centre for Stem Cell Biology at the University of Sheffield, says the MIT team?s key advance is suppressing the cell?s immune response to RNA. He calls the work an interesting approach, but adds ?the jury?s out? on whether it will prove better than other methods. ?The next step would be to make iPS cells [induced pluripotent stem cells]? using this technique, says Andrews. The MIT researchers agree that determining whether this will work remains an open question.

Explained: Bandgap

Why do some materials work well for making solar cells or light-emitting diodes (LEDs), while other materials don?t? One key factor is having the right bandgap.

In a nutshell, bandgaps have to do with how electrons behave and what it takes to get them excited. Electrons are the subatomic particles that carry a negative charge, and that surround the nucleus of an atom. When a bunch of electrons all move together in the same direction, they form an electric current.

Electrons in an atom can be thought of as being somewhere in an array of possible ?states? ? which include their energy level, momentum and spin ? with different probabilities of being in a given state. Two electrons can?t be in the same state at the same time ? that is, at least one of these variables must differ. Some particular states are possible, and some are forbidden by the laws of quantum mechanics. Sets of possible states form regions that are called bands. Sets of states that are not possible form regions between those bands, and these are called bandgaps.

The bands closest to the atomic nucleus, called core levels, and the furthest band from the nucleus that has electrons in it, called the valence band, all keep their electrons tightly in place. The next band out from that is called the ?conduction band,? and there, the electrons are free to roam around freely.

In some materials, called metals, a valence band and a conduction band overlap, and electricity flows freely and easily through them. In other materials, called insulators, there is a wide gap between the valence band and the conduction band, making it almost impossible for an electron to get excited enough to jump from one to the other, so they block the flow of electricity.

There?s a third category, and that?s where the most interesting stuff happens. These are materials that have a narrower gap between the two bands, and they are called semiconductors. Sometimes they can act like metals, sometimes they can act like insulators, and sometimes they can have properties in between. When first discovered, they were considered useless because of their erratic, variable behavior ? until physicists figured out the mystery of the bandgap.

?It was the idea of the bandgap that allowed people to understand and harness semiconductors for optoelectronic devices? ? that is, devices that work with light and electricity ? says Tonio Buonassisi, MIT?s SMA Assistant Professor of Mechanical Engineering and Manufacturing.

When electrons get excited (by getting heated, or by being hit with a particle of light, known as a photon), they can jump across the gap. If an electron in a crystal gets hit by a photon that has enough energy, it can get excited enough to jump from the valence band to the conduction band, where it is free to form part of an electric current. That?s what happens when light strikes a solar cell, producing a flow of electrons.

Silicon, a semiconductor, is the material of choice for solar cells in large part because of its bandgap. Silicon?s bandgap is just wide enough so that electrons can easily cross it once they are hit by photons of visible light.

The same process also works in reverse. When electricity passes through a semiconductor, it can emit a photon, whose color is determined by the material?s bandgap. That?s the basis for light-emitting diodes, which are increasingly being used for displays and computer screens, and are seen as the ultimate low-power light bulbs.

?A win-win across the board?

On a recent summer morning, Maria Isabel Brum took a break from her job as an MIT custodian to read the dictionary. Originally from Portugal, the avid gardener had just learned the word ?perennial,? and she was excited to use it to talk about the watercress in her garden that had been invaded by red beetles. She looked up the word in the lexicon and spelled it out loud to Tsering Mulug-Labrang, an MIT custodian born in India, who repeated the word before writing it in her notebook.

Brum and Mulug-Labrang have gathered around a table in MIT?s Department of Facilities nearly every Tuesday for more than a year to practice their English in a class led by retired MIT employee Ellen Stordy. The advanced conversation class is part of a volunteer-based pilot program sponsored by the 1,400-member MIT Women?s League. The program offers free English as a Second Language (ESL) courses to MIT?s custodial and grounds-service employees whose primary languages include Portuguese, Spanish, Chinese, Creole and Tibetan.

In addition to four weekly classes that cover a range of skill levels, the ESL program offers free one-on-one weekly tutoring sessions. This summer, 20 students are enrolled in the program, and more than 40 members of the MIT community are volunteering as instructors, substitute teachers and tutors. The program, which has no funding, has had tremendous success to date, with supervisors from the Department of Facilities reporting that employees enrolled in the classes are more productive and confident.

?I can answer questions without hesitation,? says Brum, who has been in Stordy?s class since the Women?s League launched the program in May 2009. ?I still find it hard to spell, but I can sing in church, and I read very well.?

Stordy, who retired from the facilities department two years ago, attributes her students? steady progress to their dedication and humility. ?They are like human sponges, eager to learn more and more,? she says of Brum, Mulug-Labrang and Fernanda Freitas, the three students who have participated in her class since the program started. Brum?s English has improved to the point that she was able to understand the medical terms doctors used during her husband?s recent hospitalization. Mulug-Labrang now helps her fourth-grade son with his homework, while Freitas, also from Portugal, is studying for the U.S. citizenship exam.

John DiFava, who as director of facilities operations and security oversees MIT?s custodial and grounds-services workers, describes the program as a ?a win-win across the board? for the Institute and its service employees. ?Not only have we improved their position in the U.S., but they can now also provide a better service to MIT because they are able to better understand the needs of our customers and are able to follow directions more closely,? he says.

Helping to achieve dreams

By giving his staff permission to attend the hour-long classes during their work shifts, DiFava has been instrumental to the program?s success, says Nancy Kelly, the administrative coordinator for MIT?s $100K entrepreneurship competition. Kelly co-founded the ESL program with Women?s League member Marlyse Lupis.

Kelly and Lupis were inspired by similar ESL programs at Stanford University and the National Cathedral in Washington, D.C. and wanted to start a program at MIT. They learned about Harvard University?s Bridge to Learning and Literacy program, which began in 1999 with English classes for 38 Harvard service workers and now offers language and computer courses for more than 550 employees. Carol Kolenik, the founder of the Bridge program, was eager to share her ?best practices? with Kelly and Lupis, such as how important it is to offer quality assessments to ensure that students are assigned to the appropriate class. ?I also told them to always keep in mind to ?pilot, pilot, pilot? ? keep it small and grow it very slowly, because you don?t want it to implode,? Kolenik recalls.

The Women?s League has heeded that advice, offering just three classes when the program first began. Although the coordinators have since added two evening classes for employees who work the night shift, they are cautious about growing the program too quickly. Eventually, they would like to expand the program to include employees from other MIT departments. They would also like to secure funding so they can offer citizenship and GED courses, which they know will attract both new and current students, including Mulug-Labrang, who dreams of becoming a nurse and for whom the GED is her ultimate ?goal and focus.?

DiFava says that the program has already advanced the careers of several MIT service employees. Recently, three Tibetan grounds workers had to take an exam in order to receive a license to operate certain equipment on campus. Thanks to the generosity of a few Women?s League tutors who helped the landscapers study for the test, each worker passed the exam and received the license, which DiFava says has helped broaden the workers? skill sets and improve their financial position. ?It?s been thrilling to see that happen,? he says. ?These people come here for the American dream. Why not help them out??

3 Questions: Nicholas Roy on deploying drones in U.S. skies

In June, the Federal Aviation Administration (FAA) agreed to expand flights of unmanned aerial vehicles, or drones, along the Texas-Mexico border for surveillance purposes. Although unmanned aircraft have been used extensively by the military in Afghanistan and Iraq, the FAA has been hesitant to issue flying rights for the pilotless vehicles in the U.S. other than on a case-by-case basis, such as for border patrol. Last year, the agency promised defense officials it would unveil a plan for regulating unmanned planes this year, and it recently opened a new lab to explore how air traffic control systems can control unmanned aircraft for civilian and law-enforcement purposes. As the FAA appears to make progress with regulating UAVs, MIT News sat down with Nicholas Roy, an associate professor in MIT?s Department of Aeronautics and Astronautics and director of the Robust Robotics Group, to discuss the challenges involved with flying unmanned planes across America.

Q. What are the advantages and disadvantages of using unmanned planes for non-military purposes?

A. There are a number of advantages of using UAVs for non-military purposes, and the advantages to a large extent depend on the application. The primary application of UAVs so far, for both military and non-military purposes, has been to act as a sensor platform. Just as an example, in addition to military and border surveillance, UAVs carrying atmospheric sensors can take targeted measurements to improve our ability to predict weather and climate conditions. One major advantage of using a UAV is that these vehicles have the potential to stay in the air much longer than manned aircraft. Secondly, UAVs come in a variety of shapes and sizes, and some UAVs are much smaller than a manned aircraft could ever be. A micro UAV (or MAV) carrying imaging sensors can be used in urban environments to provide police and first responders with accurate situational awareness in the event of a disaster scenario. Another major advantage of using small-scale MAVs is that they can operate in confined spaces, such as indoors or in the urban canyon, providing information to people outside the area in a way that manned aircraft and ground vehicles cannot currently.

There are also a number of disadvantages to UAVs currently. First, from an engineering perspective, a human pilot currently gives the vehicle an enormous edge in capability. Autonomous controllers and autopilots continue to grow in sophistication, but this technology is still limited compared to what human pilots can make an air vehicle do, and limited in terms of the conditions under which an autopilot can fly. Secondly, UAVs rely very heavily on external infrastructure, such as GPS and tight network connectivity to a ground station. As a result, UAVs are very good at projecting information and situational awareness back to ground stations, but at a substantial cost in terms of human support.

Q. Why is it so tricky to regulate UAVs for non-military use? What are the technological challenges involved, and how does the FAA plan to address these?

A. When discussing large-scale UAVs (i.e., not human-portable MAVs), there is a logistical challenge of integrating the UAV not only into the airspace, but also into the infrastructure currently supporting commercial and general aviation. UAVs must be able to interact with air-traffic control and the pilots of other aircraft. This requires either substantial advances in human-robot interaction, such as speech recognition, or modification of how aircraft communicate with each other and the ground. Additionally, regulations such as right-of-way rules designed to ensure collision avoidance, also known as deconfliction, create technological challenges of developing sensors and sensor-processing algorithms that can perceive other aircraft at specific distances and sizes.

Finally, ensuring safety and minimizing risk is a challenge for both military and non-military vehicles. Air vehicles of all shapes and sizes have always been extremely safe but are becoming increasingly complex in almost every measure such as design, capability, number of components, scope of operation, etc. The growth in complexity of the entire system poses challenges for us as engineers in terms of being able to test a system, identify potential faults and certify the vehicle. Ensuring the safe operation of an autonomous perception and control system that is required to fly a UAV in unstructured environments is a challenge for all autonomous vehicles.

Q. The FAA is expected to move from a radar-based traffic control system to a GPS-based one within the next decade. How exactly will this make it easier to track UAVs and prevent collisions?

A. GPS does not necessarily make it easier to track UAVs ? there has been enormous effort in developing and deploying the satellite infrastructure, and each vehicle will require the appropriate equipment not only to track its position using GPS but also to broadcast that estimate. The expectation is that GPS will provide faster, more accurate and more precise measurements of each vehicle in the air. These improvements should allow for faster air traffic control decisions and may increase the number of available routes for aircraft to fly, which may in turn reduce the effort required to avoid collisions.

However, the plans for GPS-based air traffic control involve more than just GPS, as the position information can be accompanied by additional data about the vehicle, including identity information. This is especially useful for UAV autopilots, which can use this additional source of digitally encoded information to build situational awareness and make local decisions about trajectories. Without GPS information from other vehicles, the demands on the sensors onboard the UAV to avoid collisions are substantial.

Safety of course is paramount, and GPS is not a magic bullet. It can be subject to errors or noisy measurements, just as radar-based systems can. Switching from one sensor modality to another requires careful planning and an awareness of its limitations.

Looks like a winner

When you vote in an election, your choice is surely not influenced by anything as superficial as a candidate?s looks, right?

Right?

New research from MIT political scientists shows that the appearances of politicians do indeed strongly influence voters ? and that people around the world have similar ideas about what a good politician looks like. While few political observers would be surprised to learn that good looks earn votes, the MIT researchers have quantified a phenomenon that is more often assumed to be true than rigorously measured.

?Ever since Aristotle, people have written about the concern that charismatic leaders who speak well and look good can sway votes even if they do not share the people?s views,? acknowledges Gabriel Lenz, an associate professor in the Department of Political Science at MIT, and a co-author of the study.

To test this idea, though, Lenz and his colleagues showed voters in the United States and India pairs of candidate photos from real election matchups in Brazil and Mexico. When asked which candidate would make a better elected official, the participants in the study, regardless of where they lived, largely selected the same candidates. Moreover, their choices corresponded closely to the outcomes of those Brazilian and Mexican races, meaning the public attribution of good looks to a candidate is a leading indicator of a campaign?s result.

?We were a little shocked that people in the United States and India so easily predicted the outcomes of elections in Mexico and Brazil based only on brief exposure to the candidates? faces,? says Lenz. ?These are all different cultures, with different political traditions and different histories.?

In the study, the researchers showed voters pairs of candidates from 122 elections in Mexico and Brazil. The participants in the study were asked which candidate would be a better elected official. Respondents in India and the United States agreed with each other about 75 percent of the time when asked which candidate seemed superior; a group of respondents in the United States and Mexico agreed with each other about 80 percent of the time.

In turn, simply knowing which candidate the participants judged to have a superior appearance allowed the researchers to correctly predict the winner in 68 percent of Mexican elections and 75 percent of some Brazilian elections. ?These are very large effects,? the authors note in the working paper, ?Looking like a Winner: Candidate Appearance and Electoral Success in New Democracies,? which will be published in the journal World Politics this fall.

Of all the politicians in the study, only six candidates in Mexico and 16 candidates in Brazil were female. (The researchers chose pairings on the basis of photo availability; the pairings do not necessarily represent the precise proportion of female candidates in all such races.) Other researchers in the field, as the paper notes, have found that female candidates fare better in surveys of looks than they do in actual elections. This disparity between survey outcomes and ballot box results, Lenz thinks, could indicate that male participants in these surveys do not want to appear prejudiced in public-opinion studies, but vote in ways that differ from their survey responses. ?We would like to study these issues more,? Lenz adds.

Lenz conducted the study along with Chappell Lawson, also an associate professor of political science at MIT, Michael Myers, a research affiliate with MIT?s Department of Political Science, and Andy Baker, a political scientist at the University of Colorado.

The paper is an ?interesting and innovative study,? writes Panu Poutvaara, an economist at the University of Helsinki who also studies the influences of candidate appearances, responding to questions by e-mail. In Poutvaara?s view, by helping to confirm the general connection between good looks and ballot-box success, the study paves the way for future research that should address precisely why voters favor good-looking candidates: ?Is it because voters either enjoy watching good-looking politicians on TV, or think that they are better in social interactions??

Lenz and his colleagues are addressing this question from a slightly different angle in additional, ongoing research. In a forthcoming study, they find that ?low-information voters? are especially likely to choose candidates based on looks. ?These are people who don?t know much about politics, but watch a lot of TV,? says Lenz. The researchers are currently writing a paper based on this latter project.

Bursting a bubble?

Understanding the processes that cause volcanic eruptions can help scientists predict how often and how violently a volcano will erupt. Although scientists have a general idea of how these processes work ? the melting of magma below the volcano causes liquid magma and gases to force their way to Earth?s surface ? eruptions happen so rarely, and often with little warning, that it can be difficult to study them in detail.

One volcano that volcanologists believe they understand fairly well is Italy?s Stromboli, which has been erupting every five to 20 minutes for thousands of years, spewing fountains of ash and magma several meters into the sky. For several decades, scientists have pretty much used one theory to explain what is causing huge amounts of gas to erupt so frequently: swimming-pool-sized bubbles that travel through a few hundred meters of molten magma before popping at the surface.

But they may be wrong, according to new research by Jenny Suckale, a graduate student in MIT?s Department of Earth, Atmospheric and Planetary Sciences (EAPS), who has developed a sophisticated computer model to simulate Stromboli?s magma flow. In a two-paper series published July 20 in The Journal of Geophysical Research, Suckale suggests that giant gas bubbles can?t be driving the Stromboli eruptions because such bubbles aren?t compatible with the basic laws of fluid dynamics, or the science of how fluids move. Instead of large bubbles that pop at the top of Stromboli?s conduits ? pipelike openings that connect the volcano?s magma chamber to the Earth?s surface ? Suckale thinks that the eruptions are caused by a spongelike plug located within the conduit, similar to a cork in a champagne bottle, that fractures every few minutes as a result of pressure created by significantly smaller bubbles.

Although all volcanoes are different ? some are driven by gas while others are driven by rising magma or interactions with water ? Suckale says that figuring out Stromboli would be ?an important step forward for volcanology? because scientists don?t really know the details of how most volcanoes function. Rethinking how Stromboli works could also shed light on the processes of volcanoes that appear to be driven by similar mechanisms as Stromboli, such as Mount Erebus in Antarctica, which has been continuously active since the 1970s.

Scaling Stromboli

Despite having a wealth of data about Stromboli, volcanologists have really only applied one model to explain Stromboli?s continuous eruptions, Suckale says. According to the so-called ?big bubble paradigm,? as magma rises to Stromboli?s surface, pressure drops, and this creates gas bubbles that merge together and can become several meters wide. Eventually, these bubbles explode at the top of the conduit.

But the problem with this theory, according to Suckale, is that it conflicts with the basic principles of fluid dynamics. Specifically, magma doesn?t have enough surface tension (created when two fluids meet) or viscosity (a measure of a fluid?s resistance) to maintain bubbles larger than a few dozen centimeters. She thinks that many researchers have assumed that bubbles inside Stromboli behave similarly to bubbles in a tank of water. ?People take lab models as an analog for the volcano, but the scale is so different, and fluid dynamics is so dependent on scale,? she explains.

To test the theory, Suckale and co-authors and EAPS professors Brad Hager and Lindy Elkins-Tanton, as well as Jean-Christophe Nave, a lecturer in MIT?s Department of Mathematics, developed a computer model of the inner volcano?s mixture of gas and magma and the bubbles that can rupture or merge. By changing certain parameters, such as scale, she discovered that it would be physically impossible for massive gas bubbles in Stromboli to survive for longer than a second because of the lack of stabilizing forces, such as surface tension and viscosity.

Suckale still believes there are gas bubbles inside Stromboli that are created by some unknown source located underneath the volcano. But she thinks these bubbles are significantly smaller ? perhaps only several centimeters thick ? and accumulate beneath a porous plug that covers part of the volcano. As the bubbles exert greater pressure on the plug, it eventually fractures, causing gas, rocks and liquid to scatter into the sky. This could explain why samples of Stromboli rock contain many tiny crystals ? because the top of Stromboli is a spongelike plug of crystals and gas bubbles that releases lots of gas every few minutes.

Kathy Cashman, a geologist at the University of Oregon, says Suckale?s modeling work ?greatly advances? volcanologists? understanding of the bubbles inside Stromboli and may also shed light on noneruptive processes in volcanoes that could also be transferring gas to the atmosphere. ?Jenny?s work sits at the boundary of these two types of gas transfer, and her modeling may help to address very fundamental issues related to volatile budgets of both the magma and the atmosphere,? Cashman says. But she cautions that Suckale?s work represents a ?first step? toward modeling a very complex system, and that future modeling efforts should address the effect that crystals may have on bubble behavior.

Suckale agrees, but for now, she is working to develop a new model to explain how she thinks the theorized Stromboli plug works, why it could cause such constant eruptions and what this might say about other volcanoes that erupt frequently.

A plane that lands like a bird

Everyone knows what it's like for an airplane to land: the slow maneuvering into an approach pattern, the long descent, and the brakes slamming on as soon as the plane touches down, which seems to just barely bring it to a rest a mile later. Birds, however, can switch from barreling forward at full speed to lightly touching down on a target as narrow as a telephone wire. Why can't an airplane be more like a bird?

MIT researchers have demonstrated a new control system that allows a foam glider with only a single motor on its tail to land on a perch, just like a pet parakeet. The work could have important implications for the design of robotic planes, greatly improving their maneuverability and potentially allowing them to recharge their batteries simply by alighting on power lines.

Birds can land so precisely because they take advantage of a complicated physical phenomenon called "stall." Even when a commercial airplane is changing altitude or banking, its wings are never more than a few degrees away from level. Within that narrow range of angles, the airflow over the plane's wings is smooth and regular, like the flow of water around a small, smooth stone in a creek bed.

A bird approaching its perch, however, will tilt its wings back at a much sharper angle. The airflow over the wings becomes turbulent, and large vortices ? whirlwinds ? form behind the wings. The effects of the vortices are hard to predict: If a plane tilts its wings back too far, it can fall out of the sky. Hence the name "stall."

The smooth airflow over the wings of a normally operating plane is well-understood mathematically; as a consequence, engineers are highly confident that a commercial airliner will respond to the pilot's commands as intended. But stall is a much more complicated phenomenon: Even the best descriptions of it are time-consuming to compute.

Reap the whirlwind

To design their control system, MIT Associate Professor Russ Tedrake, a member of the Computer Science and Artificial Intelligence Laboratory, and Rick Cory, a PhD student in Tedrake's lab who defended his dissertation this spring, first developed their own mathematical model of a glider in stall. For a range of launch conditions, they used the model to calculate sequences of instructions intended to guide the glider to its perch. "It gets this nominal trajectory," Cory explains. "It says, 'If this is a perfect model, this is how it should fly.'" But, he adds, "because the model is not perfect, if you play out that same solution, it completely misses."

So Cory and Tedrake also developed a set of error-correction controls that could nudge the glider back onto its trajectory when location sensors determined that it had deviated from it. By using innovative techniques developed at MIT's Laboratory for Information and Decision Systems, they were able to precisely calculate the degree of deviation that the controls could compensate for. The addition of the error-correction controls makes a trajectory look like a tube snaking through space: The center of the tube is the trajectory calculated using Cory and Tedrake's model; the radius of the tube describes the tolerance of the error-correction controls.

The control system ends up being, effectively, a bunch of tubes pressed together like a fistful of straws. If the glider goes so far off course that it leaves one tube, it will still find itself in another. Once the glider is launched, it just keeps checking its position and executing the command that corresponds to the tube in which it finds itself. The design of the system earned Cory Boeing?s 2010 Engineering Student of the Year Award.

The measure of air resistance against a body in flight is known as the "drag coefficient." A cruising plane tries to minimize its drag coefficient, but when it's trying to slow down, it tilts its wings back in order to increase drag. Ordinarily, it can't tilt back too far, for fear of stall. But because Cory and Tedrake's control system takes advantage of stall, the glider, when it's landing, has a drag coefficient that's four to five times that of other aerial vehicles.

From spy planes to fairies

For some time, the U.S. Air Force has been interested in the possibility of unmanned aerial vehicles that could land in confined spaces and has been funding and monitoring research in the area. "What Russ and Rick and their team is doing is unique," says Gregory Reich of the Air Force Research Laboratory. "I don't think anyone else is addressing the flight control problem in nearly as much detail." Reich points out, however, that in their experiments, Cory and Tedrake used data from wall-mounted cameras to gauge the glider's position, and the control algorithms ran on a computer on the ground, which transmitted instructions to the glider. "The computational power that you may have on board a vehicle of this size is really, really limited," Reich says. Even though the MIT researchers' course correction algorithms are simple, they may not be simple enough.

Tedrake believes, however, that computer processors powerful enough to handle his and Cory's control algorithms are only a few years off. In the meantime, his lab has already begun to address the problem of moving the glider's location sensors onboard, and although Cory will be moving to California to take a job researching advanced robotics techniques for Disney, he hopes to continue collaborating with Tedrake. "I visited the air force, and I visited Disney, and they actually have a lot in common," Cory says. "The air force wants an airplane that can land on a power line, and Disney wants a flying Tinker Bell that can land on a lantern. But the technology's similar."