Why one researcher thinks stoplights may not be necessary
Most drivers have had this moment: you roll up to an empty intersection, the light turns red and you sit there while absolutely nothing comes from the other direction. No headlights. No cross traffic. Just your car, your patience and a small mechanical decision made by a box above the road.
That everyday annoyance sits at the center of Christos Cassandras’s work. Cassandras, a Boston University engineer recognized by the IEEE Intelligent Transportation Systems Society for research tied to autonomous driving, has spent years looking at a simple question with unusually large consequences: if vehicles can coordinate safely, do we need to rely so heavily on traffic lights?
The idea sounds almost cheeky at first. People are used to signals, timers and painted lane markings telling everyone when to stop and go. Strip that away, though and a different picture appears. In theory, manage its own order, a road full of connected cars could. One vehicle could slow a little, another could pass and a merge could happen without the usual stop-and-start choreography that eats up time and fuel. The point isn’t to let cars run wild. It’s to replace blunt signals with communication that’s more precise.
That matters because the cost of a needless red light’s more than a few annoyed drivers. Every unnecessary stop burns energy. It also creates delays that ripple outward, especially when traffic is already dense. If autonomous vehicles and connected cars can negotiate movement cleanly, the result could be smoother flow, fewer hard braking events and less wasted motion at intersections that currently spend long stretches waiting for nobody in particular.
Cassandras isn’t pitching a science-fiction stunt where streets suddenly become driverless free-for-alls. He is working from a long engineering tradition that asks how machines, and now vehicles, can make decisions together under rules that keep people safe. That distinction matters. There’s a big gap between a neat demo and a system that can handle real roads, impatient drivers, bad weather and the occasional person who treats a stop line like a suggestion.
His research sits in that practical middle ground. It asks what happens when autonomous vehicles coordinate rather than simply react. And it asks how much delay can be removed without making intersections less safe. And it asks whether traffic lights, which have shaped city driving for more than a century, may eventually become less central in places where cars can communicate directly.
That’s the larger thread running through this article. The argument isn’t that signals disappear tomorrow. It’s that an engineer’s spent years building the case that, in the right conditions, vehicles may be able to do part of that job themselves. It helps to start with the basic frustration: a red light, an empty road and the suspicion that there might be a smarter way, before getting to the math and the career that led there.
From Athens to engineering: the path that shaped his thinking
Christos Cassandras didn’t come up through the usual physics-only pipeline. He grew up in Athens during Greece’s military dictatorship, a political setting that could hardly have encouraged easy answers or blind trust in institutions. That kind of environment tends to make a person notice how systems work, who controls them, and what happens when they fail. For a young student with broad interests, that curiosity mattered. Cassandras was drawn to more than equations. Philosophy, history, science and engineering all had a place in the conversations around him and that mix pushed his thinking toward questions about how people make choices inside larger systems.
His path makes more sense once you see that he was never interested in a single discipline for its own sake. He wanted to know how ideas, decisions, and machines fit together.
Two moments from that era seem to have landed hard. The Apollo 11 moon landing showed what coordinated engineering could do when a huge technical effort had to work in the real world, not just on paper. Which helped open the modern networked age, pointed in a different direction: machines could exchange information and respond to one another, even across distance, given the first computer-to-computer data transfer. For a teenager already thinking about structure and decision-making, that must’ve felt like a door opening. The future wasn’t just about bigger engines or cleaner formulas. It was about connected systems that could react.
A scholarship took him to the United States, where he began college at Yale. He first moved through physics and philosophy, which makes sense for someone trying to understand both the material world and the logic people use to describe it. In time, though, engineering pulled him in more strongly. That shift wasn’t a rejection of the earlier subjects. It was more like a practical turn toward the place where abstract ideas meet machinery, control and real constraints. Later graduate work at Stanford and Harvard gave him even more room to study how systems behave over time, especially when many parts are acting at once.
Game theory was one of his early interests, and it’s easy to see why. It deals with planned behavior, tradeoffs and choices made under pressure. But a mentor redirected him toward active systems and discrete-event thinking, an area that was still forming into a distinct research path. That nudge mattered. Instead of focusing only on planned moves between rational players, Cassandras began to study systems that change when events happen: a sensor fires, a packet arrives, a machine finishes a task, a vehicle slows down. The details may sound technical, but the idea’s plain enough. Some problems don’t unfold smoothly. They jump.
That shift from strategy to system behavior would later shape work in smart transportation and connected vehicles, where timing, coordination and response matter more than theory alone. For now, though, the important part’s how early his outlook took shape. Athens gave him a political and intellectual backdrop. Yale, Stanford and Harvard gave him the tools. Game theory gave him one way of thinking. Active systems gave him a better one for the problems he would spend years studying.
The Fiat project that helped launch discrete-event systems
The move from abstract theory to something you can point at happened, as these stories often do, with an assignment that sounded ordinary on paper. Cassandras was asked to work on a Fiat assembly-line buffer problem, the kind of task that can look like routine manufacturing bookkeeping until you actually try to model it. Suddenly, it wasn’t just about parts moving through a plant. It was about what happens when a buffer fills up, when a station runs dry, and when one small delay ripples through the whole line.
That was the moment the logic of event-driven systems started to feel real. In classical physics, you usually track things that evolve continuously. Position changes every instant. Speed changes every instant. Temperature, force, motion, all of it’s in motion whether you stare at it or not. A manufacturing line’s different. Most of the time, it sits there. Then a part arrives. Fair enough. A machine finishes a task. And a queue crosses a threshold, a button gets pressed. Something happens, and the system reacts.
Some systems live by the clock. Others sit quietly until a part shows up, a signal arrives, or a switch flips.
Naturally, that distinction is the heart of discrete event systems. The name sounds a bit dry, like something scribbled on a whiteboard after too much coffee, but the idea’s practical: model systems around the events that matter rather than pretending everything changes smoothly all the time. In a factory, that means the arrival of a component may matter more than the second-by-second motion of a conveyor belt. In a control room, the moment a sensor trips can matter more than the steady background state of the machine. Once you start looking that way, a lot of engineered systems stop resembling textbook physics problems and start looking like sequences of triggers, waits and reactions.
Cassandras has said in different settings that this was a natural extension of the questions that had already pulled him toward decision-making and systems. The Fiat project gave those ideas a concrete shape (for better or worse). It also pointed him toward later work on hybrid systems, where continuous behavior and discrete events live in the same model and then toward cyber physical systems, where software, sensors and physical processes are tightly mixed. A traffic signal, a robot arm, a vehicle braking because a sensor detected an obstacle, all of them sit somewhere in that territory.
He didn’t treat the academic side as separate from the practical side for long. After that early work, he spent a short stretch at a Boston-area startup focused on manufacturing automation. It was a brief industry stint, but it mattered. He kept seeing how models behaved when they met messy shop floors, real deadlines and equipment that didn’t care how elegant the equations looked. That habit of staying close to industry never really left. It shows up in the topics he chose, the systems he studied and the way he talks about engineering problems as things that have to work in the world, not just on paper.
His academic path later took him to the University of Massachusetts Amherst, where he joined the faculty and built on that systems focus. Where he helped build a systems engineering division and later led its research program, after that came Boston University. By then, the thread running through his work was fairly clear: treat engineered systems as systems, not just as isolated machines or tidy equations. Look for the events, the triggers, the waits, the bottlenecks. That lens turned out to be useful far beyond a Fiat factory floor, and the next step was to see how it could help cars talk to one another before anyone trusted them to share a road.
Safety first: proving autonomous vehicles can coordinate
Coming off the early engineering work, Cassandras’s current research gets even more concrete. He and his team are trying to answer a simple but unforgiving question: can a connected vehicle prove its move’s safe before it makes it?
That sounds tidy on paper. In practice, it means building mathematical models from things a car can actually measure or estimate, like position, speed, mass and the distance to nearby vehicles. Feed those numbers into the model, and the software can work out whether a merge, a lane change, or some other maneuver leaves enough room for everyone to get through cleanly. The idea’s especially useful for connected and automated cars, where vehicles can share information and coordinate their actions instead of each driver, or onboard system, guessing in isolation.
Safety has to be computed before it is trusted.
That’s the basic logic behind a lot of his work. If a car can check a maneuver against a safety model first, it can avoid making decisions that depend on optimism, luck, or a last-second apology brake tap. In theory, that makes self-driving car safety less about hoping the machine behaves and more about proving the behavior’s acceptable before the tires roll.
One of the better-known demonstrations came from Boston, near Commonwealth Avenue and the Boston University Bridge. There, the team studied what would happen if the traffic lights were removed and vehicles coordinated among themselves instead. That’s a bold idea, but not a reckless one. The point wasn’t to throw out traffic control and let chaos sort itself out. It was to run simulations and see whether communication between vehicles could manage the same intersection more smoothly than a fixed signal cycle.
The reported results were promising. Traffic flowed better, safety margins improved and energy use dropped because cars spent less time idling at red lights or lurching into stop-and-go patterns. Anyone who has sat through a green light that turned red after one car crossed the line knows the emotional appeal of that setup. Still, the gap between simulation and city streets remains large. Only a small fraction of cars can communicate with one another right now, so fully signal-free intersections are more of a research target than a commuter reality.
That limitation matters. A system built for fully connected vehicles can look elegant in the lab and awkward on a road filled with ordinary cars, delivery vans, cyclists and buses that all behave a little differently. Cassandras has spent enough time around real systems to know that neat theory gets messy once humans enter the picture. So his work tends to stay grounded in what can be measured, tested, and adjusted rather than what merely sounds clever at a conference.
He has also used small robots in his lab to test coordination ideas at a scale that’s easier to control than a street full of cars. That may sound playful, and to a degree it is. Tiny robots are cheaper than wrecked fenders. They also let researchers watch coordinated motion unfold in a controlled environment, where one bad assumption doesn’t turn into a police report. The lessons from those experiments can then be carried into larger systems, where the stakes are higher and the margins are thinner.
Across that research, a pattern shows up again and again. Cassandras seems less interested in technology that impresses other engineers than in technology that works around actual human behavior. That shift matters. A model can be mathematically elegant and still be useless if it ignores how people drive, hesitate, cut in, or simply don’t own the newest connected hardware. At Boston University, that practical lens has shaped much of his recent work, which asks not just whether autonomous systems can move, but whether they can move safely in the world people actually use.
A career built on students, IEEE, and practical impact
One way to tell whether a research program’s legs is to look at where the students end up. D. True enough. Students have moved into safety-related work at companies such as Aptiv and Zoox, where the questions are no longer academic puzzles but live engineering problems: how a vehicle decides, how it reacts, how it avoids doing something expensive or dangerous in a split second. That kind of movement from lab to industry usually says more than a polished bio ever could.
He has also tried to turn abstract math into something engineers can actually use. His coauthored book on safe autonomy and control barrier functions does that in a fairly direct way. Instead of treating safety as a vague promise, the book lays out a mathematical foundation for keeping systems inside safe operating limits. That matters because elegant theory can sit on a shelf for years if nobody knows how to plug it into a real control problem. Here, the writing appears aimed at making the method usable, not just admirable.
Useful autonomy has to survive contact with traffic, weather, hardware limits, and human behavior. Otherwise, it’s just a neat simulation.
His IEEE involvement runs almost as long as his academic one. He started as a student member, then stayed involved through the years in leadership and committee roles, including work connected to the IEEE Intelligent Transportation Systems Society. That sort of long-term participation can sound dry on paper, yet it often shapes what gets discussed, funded and shared across the field. Professional societies move slowly, but they matter because they create a place where people with different specialties can compare notes before everyone rushes off to build the next system.
The through line in all of this is pretty plain. Cassandras seems less interested in technology as a trophy than in technology as a set of rules, tools and machines that people can live with. If a model helps a car merge safely, if a student carries that thinking into a company building autonomous systems, if a book gives engineers a way to formalize safety instead of waving at it from a distance, then the work’s done something useful. Not flashy. Just useful.
That may be the cleanest way to understand his career. It’s room for math, robots, committees, books, plus students, but none of it floats free of the real world. The point, as he puts it through his work, is to make everyday systems safer and easier for the people who rely on them, whether they’re sitting behind a steering wheel or crossing the street with a coffee in one hand and a phone in the other.
