Printer-friendly version

Professor Edward A. Lee is co-author with Berkeley colleague Pravin Varaya of the book Structure and Interpretation of Signals and Systems (Addison Wesley, 2003) Cordell Green photo

Imagine if on September 11, 2001, New York City was surrounded by a force field. When the terrorists flying toward the World Trade Center began their descent, they would have encountered a phenomenon not unlike a vortex that pushed the plane to the left or right. As they fought toward their target, the resistance could have increased until the plane was automatically diverted from lower Manhattan.

This kind of virtual bubble around "forbidden zones" of airspace is the aim of Soft Walls, a project underway within UC Berkeley's Center for Hybrid and Embedded Software Systems (CHESS), part of the Center for Information Technology Research in the Interest of Society (CITRIS).

"We're trying to solve the problem that aircraft can be used as weapons," says Professor Edward A. Lee of the Department of Electrical Engineering and Computer Sciences (EECS). Lee is collaborating on the project with EECS chair Shankar Sastry, graduate students Adam Cataldo, and postdoctoral researcher Ian Mitchell.

The Soft Walls system would be embedded in new aircraft and does not depend on any air traffic control infrastructure or networking technology. The approach takes advantage of modern aircrafts' "fly-by-wire" system that translates a pilot's commands into the computer instructions that actually control the aircraft. The Soft Walls software, Lee explains, contains a database of "no-fly zones." Using a plane's existing gyroscope-based inertial navigation system and Global Positioning System (GPS) technology, the on-board computer checks the plane's location against the database.

This diagram depicts several responses of the Soft Walls system on a plane about to enter a no-fly zone. Courtesy Adam Cataldo

"The idea is to constrain the airspace within which an aircraft can fly while maintaining the maximum amount of pilot authority," Lee says.

If the plane is heading into a forbidden zone, the pilot will first be notified visually and resistance will build. If the pilot does not cooperate and change the flight path, "the controls will eventually saturate and the aircraft will be diverted," Lee explains.

Even at that point, Lee adds, the pilot will maintain fine-grain control of the aircraft to avoid dangers like a mid-air collision with another aircraft that may be in the no-fly zone.

"We have viable control algorithms and a strategy for figuring out how to blend the pilot's input with the control system," he says. "But there's a lot more work to ensure that the solution is robust."

For example, the security of such a system is a huge concern. Jamming the aircraft's GPS so it can't calculate its location or "spoofing" the system into thinking it's somewhere else would be catastrophic, Lee says. The researchers are currently collaborating with experts in flight navigation systems to identify and block any potential security flaws.

Once the technical challenges are ironed out, will Soft Walls fly? Boeing and Honeywell are interested, Lee says. Meanwhile, NASA is leading a project to build consensus in the entire aviation industry about the most effective and acceptable method to prevent airplanes from becoming missiles. The toughest sell on Soft Walls, Lee says, are pilots.

"There's a 2,000 year-old tradition of a ship's captain that gives a pilot tremendous authority on board a craft," he says. "There's a lot of suspicion in aviation of any technique that attempts to limit that authority in any way."

The control specimen was not reinforced with the polymer mesh. One dominant crack is visible. The specimen had low strength and low toughness.

An ingeniously simple and inexpensive building-reinforcement system developed at UC Berkeley could dramatically reduce the death toll in major earthquakes like those that recently rocked Turkey and India.

Civil Engineering associate professor Claudia Ostertag and her students recently tested a construction technique that prevents the low-cost adobe or brick buildings common in developing nations' rural areas from collapsing after a quake. The magic lies in the cheap and easily-obtainable materials used to reinforce the walls: In India, it's the burlap that coffee bean sacks are made from; in Turkey, the felt-like polymer used to line seat cushions is ideal.

"An adobe structure fails when one dominant crack propagates through it," Ostertag says. "So we are applying concepts from fracture mechanics to modify the crack paths in the walls."

Professor Claudia Ostertag and one of her students take a close look at a concrete sample about to be compressed to failure in a "split in tension" device. Bart Nagel photo

In adobe walls, cracks propagate vertically through each row of bricks. According to Ostertag, engineers have attempted to strengthen every component of the walls, from the bricks to the mortar to the bond between the two.

"Nothing ever worked," she says. "So we decided to move in a completely different direction. Strength is important, but not that important. What you really want is ductility, the ability to absorb energy."

The goal is to eliminate the single dominating crack, blocking a fracture from propagating all the way up the wall. Even with a multitude of small cracks, "the wall will hold together and the building won't collapse on the occupants," Ostertag says.

The best barrier to crack propagation is fiber reinforcement, specifically a mesh fabric. When strips of the material are used to line each row of bricks and sandwich the mortar, a crack may initiate but can't penetrate through the reinforced layer. The impact this extra step has on construction time and labor is minimal.

The specimen with the reinforced mortar joints visible.

To test the reinforcement technique, Ostertag and her students built a full-scale "specimen" adobe wall. The wall was then subjected to massive force from a hydraulic jack while sensors measured how well the structure held up to the pressure. In an instant, the team had proven their theory.

The challenge that remained was finding a suitable material that is cheap and readily available in each developing nation. Fortunately, they didn't have to search long.

The reinforced specimen suffered multiple cracks during testing, rather than a single dominant crack. (Blue lines indicate cracks.) These cracks absorb energy and will not cause the wall to collapse.

"One student visiting El Salvador asked me what he should look for," Ostertag says. "I told him to visit the local market and see what they carry their produce in. The bags turned out to be a polymer material."

After the success of the adobe wall experiment, Ostertag, in collaboration with assistant professor Khalid Mosalam, conducted a similar test in the civil engineering test bay using fired masonry bricks instead of adobe and strips of a high-strength polymer mesh. Again, tiny cracks appeared but the structure did not fail.

The next step, Ostertag says, is to construct an entire mesh-reinforced building on UC Berkeley's "shake table," the nation's largest earthquake simulator. Located at the university's Richmond Field Station, the twenty-by-twenty foot table is capable of three-degrees of hydraulic motion to accurately replicate seismic activity. The researchers are currently seeking funds to support the shake tests.

While Ostertag's efforts are aimed at saving lives in developing nations, the same approach could protect buildings in industrialized nations, she says.

"The death count in major earthquakes around the world is unacceptable," Ostertag says. "People don't need to live in unsafe buildings."

A snapshot of the ShareCam interface with a view of the Stanley Hall construction site. Courtesy the researchers

Nobody wants to wait in line. This is especially true on the World Wide Web, where myriad distractions are just a mouse-click away. UC Berkeley professor Ken Goldberg and graduate student Dezhen Song are tackling this problem with regard to telerobotic webcams, Internet-connected video cameras that users currently queue up to control. ShareCam, the researchers' collaboratively-operated robotic webcam system, eliminates the wait.

"We wanted to create a way for a robotic webcam to be controlled by many people at once," says Goldberg, who holds a joint faculty position in the Department of Industrial Engineering and Operations Research and the Department of Electrical Engineering and Computer Sciences.

Commercially-available and inexpensive webcams with pan, tilt, and zoom controls are being installed in hundreds of locations, from Hiroshima Harbor to the Giant Panda habitat at the National Zoo. The conundrum, Goldberg explains, is that the more popular a webcam, the fewer number of people get to use it.

"When there's a lot of demand, robotic webcam systems simply don't scale," he says. On the other hand, ShareCam could potentially scale to millions of simultaneous users, Goldberg adds.

To participate in the collaborative control of ShareCam, users register online and select a color to represent themselves. The ShareCam interface contains two windows. The first provides a streaming video view from the camera's lens. The second window depicts a static image of the camera's entire range of vision. A user simply draws a box, or frame, around the portion of the image he'd like the camera to hone in on. The size of the frame characterizes the zoom of the camera. Meanwhile, nineteen other users do the same thing. The computer then tallies the user requests and calculates an "optimal" single frame. The camera then shifts its gaze to that frame.

Ken Goldberg is the editor of two books on Internet telerobotics, both published by MIT Press: The Robot in the Garden and (with Roland Siegwart) Beyond Webcams. Bart Nagel photo

The magic of the ShareCam software is its ability to please most of the people, all of the time. A seemingly obvious approach would be to average all of the votes to determine the camera's next position. Failure quickly occurs though, Goldberg says.

"If you have a bunch of votes to move right and a bunch to move left, the camera would end up moving to the middle of the picture, which nobody wants," he says. "So we needed a more sophisticated model that resolves conflict between all the different votes."

Goldberg and Song came up with a novel satisfaction metric, a method to measure the satisfaction of each user with respect to the potential camera frame computed from the votes. The researchers then devised new algorithms that enable the system to find a camera frame that optimizes the measure of total user satisfaction.

"The algorithms are very fast and very computationally efficient," Goldberg says.

The ShareCam research ties directly to Goldberg's online experiments in collaborative filtering, the notion that people who agreed in the past will probably agree in the future. The project also follows Goldberg and his team's development of the TeleActor, a system where users democratically control a human "robot" to explore remote spaces. Like the TeleActor, the ShareCam project is part of a larger distance learning effort in the Berkeley-based Center for Information Technology Research in the Interest of Society (CITRIS).

A prototype ShareCam is now mounted outside of the new Stanley Hall construction site. When completed, the Stanley Hall Biosciences and Bioengineering Facility will house laboratories for CITRIS, the California Institute for Quantitative Biomedical Research (QB3), and the Department of Bioengineering.

"We want to place cameras where there's a lot of public interest," Goldberg says. "We've also talked about locations like a live volcano or in a war zone — anywhere lots of people would like to see but it's not practical to install very many cameras."

For over forty years, the Microfabrication Laboratory has been the site of innovative and cutting-edge research. Bart Nagel photo

From microcircuits and MEMS to bioengineering and nanotechnology, the UC Berkeley Microfabrication Laboratory in Cory Hall is the hub of some of the College of Engineering's most innovative research efforts, as well as supporting research across the entire campus. The University's pioneering microelectronics research and instruction began at the dawn of the digital age, just a few years after the invention of the silicon chip.

With an H6 hazard classification, the new Microlab will safely accommodate a wide variety of gases, chemicals, and processes, ensuring the facility's future flexibility as a laboratory for many disciplines — from electrical engineering and computer science to materials science, bioengineering, chemistry, and physics.

It was in 1962 that the first 1,200 square-foot laboratory opened its doors and researchers produced the first 3/4-inch diameter silicon wafers. The Microlab itself was part of the experiment, an effort to demonstrate the feasibility of integrated circuit research in a university environment.

In the 1960s, the Microlab not only proved itself an educational success, but helped drive the emerging semiconductor industry forward. Research led by professor Donald O. Pederson — whose lobbying was instrumental in making the Microlab a reality to begin with — led to the Simulation Program with Integrated Circuit Emphasis (SPICE), a tool for circuit design. SPICE or one of its myriad derivatives has been wielded in the design of every integrated circuit developed in the last 25 years.

Microlab research in the 1970s blazed trails in Metal Oxide Semiconductor (MOS) devices and spawned now-ubiquitous circuits such as switched capacitor filters and the analog-to-digital and digital-to-analog converter circuits used in today's mobile phones. As the semiconductor industry continued to gain steam, UC Berkeley kept pace. Construction on a new adjacent microfabrication facility began in 1981 while work continued in the original laboratory. In 1983, the two sections were joined and the present 12,000 square foot microlab went into operation. Nearly half of that space is devoted to a clean room where the air quality, temperature, and humidity are regulated to protect the sensitive equipment and processes.

While research pushes onward in the Cory Hall facility, the Microlab team is preparing to make history yet again with a new laboratory. The new Microlab, boasting an 18,000-foot two-story clean room, will be integrated within a building slated for construction to house the Center for Information Technology Research in the Interest of Society (CITRIS). With an H6 hazard classification, the new Microlab will safely accommodate a wide variety of gases, chemicals, and processes, ensuring the facility's future flexibility. The lower story of the Microlab will exceed the highest-vibration control standard, enabling the employment of the most delicate fabrication techniques. The current facility in Cory Hall will continue operating until the new Microlab is completed, estimated for the third quarter of 2006.

Whether the goal is a revolutionary microprocessor architecture like VIRAM or micromechanical marvel like Smart Dust, researchers in UC Berkeley's Microfabrication Laboratory not only keep up with silicon technology's state-of-the-art, they define it.

Comments, questions, suggestions?
Send us your feedback by emailing lab-notes@coe.berkeley.edu.

May 2003

1944: Metallurgist Earl Randall Parker joins the UC Berkeley Engineering faculty I have a decidedly personal story regarding Professor Parker. In 1978, I was his and Professor Zackay's secretary. One day he asked me why I was a secretary and whether I had ever considered going back to school. I replied that indeed I was considering going back to school, maybe to study chemistry or physics. He replied, why not engineering? I told him I thought I wasn't intelligent enough for engineering. Then he said the words that changed my life: you don't have to be intelligent to be an engineer — you just have to work hard. I went on to study engineering, first at Diablo Valley College part-time (while still working full-time for Professors Parker and Zackay) and then I went full-time to Berkeley. Professor Parker suggested a double major would be better than "just" a degree in Materials Science, so I took a double major in Mechanical Engineering and Materials Science. Professor Parker appointed himself my surrogate father — he continually checked on my grades, allowed me to do my homework at my desk, and "borrowed" graduate students to help me with my homework. I went on to be awarded the Departmental Citation for Materials Science in 1982 (my name's on the plaque in the Department Office, or it was last time I looked!). I went on for a Masters Degree in Materials Science and now work at LLNL as the Deputy Division Leader for the Proliferation Detection and Defense Systems Program. All because Professor Parker said that all I had to do was to work hard! Seriously, though, it was Professor Parker's continued support and encouragement that allowed me to succeed. — Jean Hodson de Pruneda BS '82, MS '85 Thanks for the write-up on Professor Parker. I greatly enjoyed studying under him in the 50's. Still remembered is the simple lab experiment he set up to show that a common volt-amp relationship assumption in welding was not true — no one had bothered to correct the old handbooks! — Ron Jameson, MetE '55

May 2003

A Shot at a New Drug-Delivery System This syringe is a great way to deliver any kind of pharmaceutical "out in the bush." It might also be a super way to deliver hormonal contraceptives. When enabling more people to survive killer diseases, it becomes more essential than ever to enable them to also space out their children and decide how many to have. It does them no good to help them survive typhoid, only to see them or their children starve to death because there is not enough food for everybody. We in the U.S. are used to having 2, maybe 3 kids in a family. Most people in developing countries would love to have families that small; currently many have 5 or even 8 or more kids per family because effective contraception is as hard for them to get as effective medicines. — Barb Parcells