Smart Car Researcher Is Reinventing Traffic

Christos Cassandras has spent greater than 40 years finding out how computer systems and machines make choices. Recently, his work has targeted on an issue practically each driver is aware of properly: sitting at a crimson gentle with no cross site visitors in sight, questioning why it takes so lengthy for the sunshine to show inexperienced.

Cassandras, an IEEE Life Fellow, is a professor at Boston University, the place he headed its systems engineering division, a graduate program he helped create.

He’s the recipient of this 12 months’s IEEE Intelligent Transportation Systems Society Outstanding Research Award for his work on autonomous vehicles and the techniques that assist them drive—work that would at some point do away with site visitors lights altogether.

That concept—proving with mathematical fashions and clever algorithms utilizing automobiles’ relative speeds, mass, and distance as inputs that self-driving cars will behave safely earlier than they’re ever let free on the streets—has develop into the inspiration of Cassandras’s profession and the explanation his title retains developing in conversations about transportation.

Within the Nineteen Nineties he theorized {that a} machine equivalent to a automobile may very well be understood as two issues directly: an object obeying the legal guidelines of physics and a pc processing info and making choices. This led to what we now name cyber bodily techniques. That framework continues to be the usual means engineers describe a automobile’s mechanical conduct.

“In the end, the important thing phrase is security,” Cassandras says. “You wouldn’t purchase a self-driving automobile until the producer might assure that it’s secure for you.”

Rising up in Greece throughout a dictatorship

Born in Athens, Cassandras was a youngster throughout Greece’s military dictatorship, a interval he says scarred him and plenty of others but in addition taught him some arduous classes. His father was a accomplice in an insulation manufacturing firm, and his mom was a homemaker.

His buddies pulled Cassandras towards engineering. As a youngster, he ran with a gaggle of classmates who examine philosophy, historical past, and science. They have been particularly struck by the 1969 Apollo 11 moon landing and the questions it raised in regards to the know-how that made it potential.

That very same 12 months additionally marked the first time a piece of data was sent from one computer to another—an early step towards the creation of the Web. A number of inspiring academics pushed his curiosity about know-how even additional.

He was awarded a scholarship to review within the United States, and he enrolled at Yale, the place he first studied physics and philosophy earlier than switching to engineering. He earned his bachelor’s diploma in engineering and utilized science in 1977. On the time, Yale didn’t but supply separate levels in fields equivalent to electrical or mechanical engineering.

From there, Cassandras went to Stanford, incomes a grasp’s diploma in electrical engineering in 1978. There, he turned serious about game theory, the mathematical research of decision-making and technique.

A mentor pointed him to Harvard, house to a number one game theory researcher on the time. His new mentor, Professor Yu-Chi “Larry” Ho, now an IEEE Life Fellow, suggested him to desert sport concept, which Ho thought of a scientific lifeless finish on the time. Ho instructed a more recent, less-explored space of analysis, that of rising dynamic techniques in trendy know-how whose conduct may very well be understood and managed via occurrences of discrete occasions.

Cassandras took the recommendation. He earned a second grasp’s diploma from Harvard the next 12 months and accomplished his Ph.D. in utilized mathematics there in 1982. He and Ho stay buddies and colleagues.

Fiat meeting line births of a brand new form of engineering

Ask individuals in intelligent transportation systems circles why Cassandras’s title carries a lot weight, and the reply traces again to a theoretical shift he helped pioneer beginning within the Nineteen Eighties. Cassandras acknowledged that human-made techniques—computer systems, manufacturing unit machines and, ultimately, autos—function on a distinct form of logic than the one which governs the pure world.

The bodily world runs on time-driven physics, the identical math Isaac Newton used centuries in the past. However machines constructed by individuals, he realized, are higher understood as event-driven: They principally sit nonetheless till one thing occurs, like a button click on or a component arriving on a conveyor belt.

That perception didn’t come from a textbook. It took form throughout Cassandras’s early graduate work, when he was handed a real-world downside. He was assigned to determine how you can handle the buffers—the non permanent holding areas for automobile elements—on an meeting line for Fiat, the Italian carmaker.

Fixing that concrete, sensible puzzle helped him see how event-driven pondering might describe a complete class of techniques that classical physics-based math couldn’t deal with properly. This got here to be often known as discrete event systems.

By the early Nineteen Nineties, that perception had matured into a proper framework: hybrid techniques concept, now often known as cyber physical systems concept. The thought is {that a} machine could be understood as an object obeying the legal guidelines of physics and as a pc obeying event-driven dynamics.

Constructing “event-driven” techniques

After finishing his doctorate, Cassandras wished to check himself exterior of academia. He spent a few 12 months and a half as a techniques engineer at Info and Expertise for Manufacturing (ITP), a small manufacturing-automation startup within the Boston space. The expertise satisfied him that the “actual world” wasn’t so completely different from tutorial life, and he returned to analysis whereas persevering with to seek the advice of for the corporate for roughly a decade. He has stored up the behavior of staying related to trade all through his profession.

In 1984 he joined the University of Massachusetts Amherst as a college member within the electrical and computer engineering department. Constructing on the concepts born out of the Fiat mission, he helped pioneer analysis into event-driven techniques.

That was a shift away from the standard math used to explain the bodily world towards techniques constructed round discrete occasions. The transition, pushed by the rise of computer systems, turned the spine of Cassandras’s analysis for the following three many years.

In 1997, after years of commuting weekly between Amherst and Boston for his consulting work, he joined Boston College as a tenured full professor in what was then its manufacturing engineering division. As the sector developed within the early 2000s with the rise of the Web and sensor know-how, Cassandras helped set up the college’s systems engineering division. In 2008 he turned the analysis program’s first director, a job he held till stepping down two months in the past.

Utilizing math to indicate that self-driving automobiles are secure

Cassandras’s latest analysis has targeted on mathematically proving vehicle safety earlier than a automobile is road-tested. He has developed fashions constructed round measurable portions: a automobile’s place, velocity, and distance to close by autos. That lets related and automatic autos cooperate throughout difficult maneuvers equivalent to altering lanes on a freeway in site visitors.

He has prolonged the method to a much bigger concept: eliminating site visitors lights. Certainly one of his extra hanging tasks entails a chronically congested intersection close to his workplace in Boston, the place Commonwealth Avenue meets the Boston University Bridge. Utilizing computer simulations, he and his group modeled what would occur in the event that they eliminated all of the site visitors lights, and automobiles merely coordinated with one another as an alternative. The outcomes, he says, confirmed enhancements not solely in how easily site visitors would circulation but in addition in security and energy efficiency.

Cassandras acknowledges that widespread adoption continues to be far off. For now, solely a small proportion of autos on the street are able to that form of communication. However he’s testing related coordination techniques utilizing small robots in his BU lab, working to deliver the thought nearer to actuality and to take advantage of the intelligence of cooperating autonomous vehicles even when they’re solely a small fraction of the particular automobiles on the street.

He says his perspective has developed in recent times from viewing people as obstacles to clean automation towards designing know-how that truly works for individuals.

By means of tasks with Boston on smart city initiatives, he discovered that not everybody can afford or entry new know-how—a lesson that reshaped how he approaches his analysis.

“It’s not nearly know-how,” he says. “It’s actually about know-how and folks.”

A legacy handed on via his college students

His affect within the subject has been amplified by a few of his college students. A number of of his Ph.D. college students have developed security algorithms at firms together with Aptiv and Zoox, each of which construct know-how for autonomous and connected vehicles. He wrote a ebook, titled Safe Autonomy with Control Barrier Functions: Theory and Applications, with one in every of his former college students. It lays out the mathematical foundations of autonomous-vehicle security, with particular purposes to clever transportation.

A protracted relationship with IEEE

Cassandras joined IEEE as a pupil member whereas he was a graduate pupil. He was drawn in by the discounted pupil convention registration charges, he says.

That early, sensible choice grew into one thing larger: an expert group with which he has stayed related for greater than 40 years.

He has served as president of the IEEE Control Systems Society, his main skilled house inside IEEE. He has served on the conference publications committee and is at present concerned with IEEE Publishing Ethics.

IEEE’s best worth to his profession, he says, has come via the relationships he has constructed at its conferences and serving on committees. Connections with colleagues and IEEE management have formed his pondering and opened doorways all through his profession, he says.

Trying again on greater than 4 many years of labor, Cassandras says the purpose has by no means actually been in regards to the know-how for its personal sake.

“In spite of everything,” he says, “we’re doing all this to make society higher and facilitate the snug lives of all of humanity.”