SE PhD Students Win TC Outstanding Student Paper Prize
Sabouni and Ahmad aim to make roads safer with event-triggered Control Barrier Functions
By Chloe Cramutola
Despite starting in different disciplines—Ehsan Sabouni in mathematics and H M Sabbir Ahmad in technology—their shared passion for intelligent transportation systems led them to collaborate on research that would later receive a prestigious award.
Their collaboration resulted in a novel paper that won the Smart Cities Technical Committee Outstanding Student Paper Prize in 2024. The paper aimed to bridge the gap between theoretical control systems and real-world traffic safety applications.
“It’s affirmation or confirmation that what we are doing is relevant to the community that we are part of,” said Ahmad, a fourth-year Systems Engineering PhD student. “And it also motivates [us] to work harder to achieve further in terms of our research goals.”
After countless hours of research and commitment, Ahmad and Sabouni’s paper, “Optimal Control of Connected Automated Vehicles with Event-Triggered Control Barrier Functions: a Test Bed for Safe Optimal Merging,” took one year to complete.
“No frustration,” Sabouni said. “Just sleepless nights.”
Sabouni is a fifth-year Systems Engineering PhD student whose research focuses on safety-critical systems. When he started working with Ahmad on their award-winning paper, Sabouni said the process took much longer because they started from scratch.
Their work was one of the first to implement event-triggered Control Barrier Function-based control on real hardware, so they had little reference, Ahmad said. Event-triggered control helps save resources by only acting when certain events happen. The CBF-based technique is used, for example, when robots operate in an open environment and on humans, where it should be ensured that the robots avoid obstacles safely, Ahmad added.
The overall goal of their research was to implement a safe control system for traffic applications.
“You want to make the road safe again by using our safe controllers,” Sabouni said. “You don’t want [autonomous agents] to collide with each other. You want them to operate in an optimal sense … without any compromise on safety.”
Ahmad and Sabouni primarily focused on autonomous vehicles, or vehicles that do not need human drivers. A main challenge in traffic networks is ‘conflict points’—areas where vehicles from different directions may intersect at the same time. Ensuring safety at these points was a major focus of their research.
“It is important to ensure that, while the cars arrive at these conflict points in the network, they do not crash into each other,” Ahmad added. “So safety is really very important, and it is critical in this system as well.”
To conduct this research successfully, the first stage in their process involved identifying suitable robots – understanding features and system models to implement the algorithms designed in Ahmad and Sabouni’s lab. The next stage involved gradually implementing the algorithms and identifying what was not working – debugging the issues, Ahmad explained.
Once implemented, the two SE students used multiple robots to “simulate a traffic system with multiple agents running seamlessly, reaching their destination without colliding into each other,” Ahmad said. If the robots run smoothly,
then that means they are running safely and optimally because they are coordinating properly to reach their destinations.
“Nothing directly translates from simulations to physical, actual hardware. So there is always that bit of effort whereby you don’t directly see any fruits or results while working on the implementation,” Ahmad said. “But over time, if you persevere, and through modifications to the actual theory, it is generally possible to make it work.”
For the two PhD students, the last phase was about successfully running these algorithms and getting the results to write in their paper.
At the core of their work was a strong motivation to improve real-world safety. “This is the beauty of working with colleagues and having a team around yourself of motivated people and people that … are really committed to finish the work,” Sabouni reflected. “And another thing is I’m really passionate about this type of work, so this also can help get me through these difficult times.”
Theory and application is a back-and-forth dance. There are limitations with hardware, and it can be too expensive to try multiple times, Sabouni said. So, he and Ahmad developed their theories and tested them out in the lab environment to decide whether the algorithm was worth implementing on actual robots.
Their work is largely based on developing novel ideas and validating them through simulations, Ahmad added. In order for these simulations to be useful or implementable in the real world, they need to validate them or implement them with existing hardware.
“As much as theory is important, it’s also important to ensure that the theories that we work on have actual real world, not only significance, but can be utilized in the real world for solving actual problems,” Ahmad said.
For both Ahmad and Sabouni, safety is the common theme in each of their respective research. Ahmad emphasized its critical role in autonomous and artificial intelligence systems.
While both PhD students said they are grateful for the award and recognition, they also said seeing their lab experiments and simulations on hardware and physical systems was enough.
“I would like to call these types of awards a cherry on top,” Sabouni said. “Because you already got what you wanted by seeing the fruit of your work, and that it actually matters and is meaningful in the real world.”
As for the future of their work, Ahmad and Sabouni said there is always space to grow and to improve their work in the lab so they can see it deployed on the street.
“Receiving the award also confirms that the work we are doing is indeed relevant and has practical usefulness in the domain that we work in,” Ahmad said. “It also motivates us further to work towards this direction, which is very closely related to both of our PhD research.”