When Dania Ammar moved with her family from Beirut to southeast Michigan in 2017, her plan was to build on her civil engineering background and start the master’s in construction management program at UM-Ann Arbor. But some family circumstances led to her deferring her enrollment. As it happened, during the unexpected semester off, she reconnected with a friend from Lebanon who was studying mechanical engineering on the Dearborn campus. He had a lot of good things to say about the college, particularly how the research enterprise was growing quickly, giving graduate students a lot of opportunities. It was enough to pique Ammar’s interest. She got on the UM-Dearborn website, checked out the programs and research labs, particularly those in Industrial and Manufacturing Systems Engineering, and set up a meeting with then-College of Engineering and Computer Science Associate Dean for Undergraduate Education and current CECS Dean Ghassan Kridli. “Dean Kridli was very generous in giving me some time to talk with him about the programs and all the research that was going on at UM-Dearborn at that time. I decided this was definitely one of the best options,” Ammar says.
The following semester, Ammar decided to go for it — enrolling in the IMSE master’s program at UM-Dearborn in 2018. IMSE is a broad discipline, and Ammar recalls feeling very open to exploring all the possibilities. That first semester, she took a variety of classes and found an early champion in Associate Professor Jian Hu, who specializes in using data analytics and machine learning in risk management applications. “I had a very good math background and knew some programming, but I had no background in that field,” Ammar says. “Professor Hu provided me with all kinds of support, told me which books I should start reading, and gave me time to get accustomed to programming models that we were using. I learned so much and it was a phenomenal experience.” A little later in her master’s program, she crossed paths with Professor Shan Bao, a specialist in human factors who was helping develop a new master’s program in Human-Centered Design and Engineering, a field that roots engineering solutions in deep understandings of human needs. Ammar says a lot of people think of human factors as a psychology-based discipline — and it does draw on psychology and user feedback quite a bit. But with Bao’s work, Ammar saw how the discipline’s deepest potential lay in combining those insights with complex data analytics and even artificial intelligence. “You can’t really develop any product, whether it is something simple or something very complicated, like an automated vehicle, without going into the human factors aspects of it,” she says. “But I saw this isn’t just about psychology. It’s about figuring out ways to quantify the psychology — to measure human preferences and behaviors so you can build models that help you gain insights into those measures.”
Ammar loved working with Bao and chose to continue her journey at Dearborn with a PhD. For her dissertation research, she took on an ambitious multiphase project in transportation safety that explored one of the more fascinating emerging challenges related to autonomous vehicles: How will pedestrians learn to trust and safely interact with vehicles when humans aren’t doing the driving? Ammar says both pedestrians and drivers typically rely on an exchange of visual signals, like eye contact and waving, to negotiate who has the right of way, and absent a human driver, cars will presumably need some way of declaring their intentions. She began her work with a deep dive into road crash data, identifying the factors that lead to the most dangerous situations between pedestrians and human-driven vehicles in order to derive potentially problematic pedestrian-AV scenarios. She then embarked on building a cueing system designed to satisfy pedestrians' needs while interacting with AVs in these evaluation scenarios. That phase of her research led to several interesting conclusions. For example, Ammar discovered that pedestrians tend to favor technologies that closely mimic their interactions with human drivers, e.g. AVs that communicate a symbolic message, like the familiar flashing silhouette used at crosswalks, either directly on the vehicle or via crosswalk infrastructure (or both). She also investigated many fine-grain details, like whether a pedestrian’s age, gender, and driving and pedestrian behavior (conservative versus nonconservative) influenced their preferences for AV cueing. The final stage of her research involved creating a VR simulation, where she observed participants reacting in tricky situations, like an autonomous vehicle quickly approaching a red light before eventually stopping. By measuring how long pedestrians took to leave the curb and using physiological measures like pupil diameter, she could see which safety cues inspired trust or distrust. She’s hoping the results, which have led to several publications, will help other researchers, automakers and civil engineers as they design cueing systems for the era of autonomous vehicles.
Since officially finishing her PhD program, Ammar has switched gears a little bit, and in some ways, it’s a moment when her diverse studies are all coming together. Since August, she’s been working as a data science project engineer with Hottinger Brüel & Kjær, where she focuses on data analytics and machine learning-based solutions for a wide variety of industries. “These skills can be leveraged in any application, including civil engineering-related fields, which have changed a lot even in the short time since I was an undergraduate student,” Ammar says. “Let’s say you had a project and you needed insights about soil mechanics or wanted to predict settlements in the soil over time. In the past, we would have done that with a bunch of physics-based and numerical equations. But now, with more sensors being installed and collected data becoming available, researchers are developing machine learning-based models to answer those questions. That’s where the field is going now, and in the future, most engineering solutions will be built with data collection in mind so these models can keep getting better.”
Ammar now just has one finishing touch to put on her UM-Dearborn career: Walking across the commencement stage in December. She expects it’s going to be a pretty emotional moment. “UM-Dearborn has been my home for the last seven years,” she says. “It’s only been a few months since I left, but to see everyone again is going to be very special. I’ve had the opportunity to work with so many different people, and I always felt, sincerely, that every day I spent with them was a chance to learn something new.”
###
Story by Lou Blouin
