Zheng Song has so many interesting projects going on it’s hard to keep up. The last time we talked with the assistant professor of computer and information science and expert on edge computing he was working on vehicle-drone interactions that could improve the response time of emergency vehicles. Earlier this year, we heard whispers of an accessibility-focused AI-based birding app that one of his Senior Design teams developed with UM-Dearborn alum and Environmental Interpretive Center staff member Donna Posont that would enable people, including blind and visually impaired folks, to bird via bird songs. The Robin app has won several awards, including at the 2025 IEEE International Conference on Systems, Man and Cybernetics, and is getting some great reviews online. Most recently, along with student Jack Wagner, president of the Artificial Intelligence Club at UM-Dearborn, Song started planning an AI4ALL seminar series, which will introduce students outside the computer science discipline to the basic concepts of AI and its benefits for everyday users. They hope to eventually expand the series to reach Dearborn high school students and local residents.
When we caught up with Song for a chat earlier this month, the subject was SpeedEye: a crowd-sourced traffic calming app that he and senior Fatima Hazime developed during the 2024 Summer Undergraduate Research Experience. Inspired by Dearborn residents’ desire to reduce speeding and reckless driving, the idea behind SpeedEye is that residents could help track which parts of the city have the biggest speeding problems, giving governments and police departments better data for informing decisions on where to deploy traffic calming and enforcement resources. To use SpeedEye, users open the app and place their phone in a window facing the street. The app uses the phone’s camera and some image recognition software to detect the models of passing vehicles. By using the length of the vehicle, the app can then estimate a vehicle’s speed. Data points are used to create a speeding heat map. Song says one of the unique things about the app is that the phone actually handles the image processing instead of sending it to the cloud for processing. Moreover, after the app uploads some basic speed and location data, images are deleted. This makes the app much more privacy-friendly, because recorded images, which might end up inadvertently including identifying information, like people’s faces or license plate numbers, never leave the phone.
In a project like SpeedEye, you’ll see all the hallmarks of Song’s recent work. First, from a technology perspective, Song is often inspired by the potential of edge computing — a decentralized approach to complex computing processes in which data processing and storage that would typically happen in large centralized locations are instead located closer to the end user. In particular, Song is drawn to finding creative ways to use consumer-grade tech that people already have in their hands. (Song envisions SpeedEye being a great use for “retired” phones that are otherwise just sitting around.) In addition, projects like this become a platform for students to do deep hands-on learning, build things and share their results with the academic community. In the case of SpeedEye, Hazime was first author of a recent paper presented at the 2025 International Conference on Computing and Machine Intelligence. And she says this research experience alongside Song has helped her clarify what she wants to do next. “Initially, I thought I’d go right into a PhD program, but now I think I’d like to work in industry for least a couple years before applying to doctoral programs,” she says. “I learned that I’m really results oriented. I want to make things that actually get into people’s hands and solve problems.” In addition to making some refinements to SpeedEye, with the hopes of launching a pilot in partnership with the City of Dearborn, Hazime will be spending her upcoming senior year working with Song on another project that’s also rooted in the needs of the Dearborn community. Through the police department contacts she made during the SpeedEye project, she learned about a VR-based driver education system that the city had rented to help train young drivers. “But this truck is expensive to rent,” she says. “We’re hoping we can build something similar for much cheaper, so cities and schools could use it whenever they want, not just three days a year.”
Perhaps most importantly, the technologies Song and his students are developing are driven by a desire to address real human needs. In the case of SpeedEye, Hazime and Song consulted with Hayat Hachem, a '21 UM-Dearborn and '23 UM-Ann Arbor alum who’s now a data scientist with the Dearborn Department of Public Health, about the particulars of the city’s speeding challenges. Hachem and Song also collaborated during a Fall 2024 web technologies course, in which students worked on a semester-long website project designed to address specific public health challenges in Dearborn. At the start of the semester, Hachem gave a guest lecture, introducing students to a wide range of issues, like flooding, speeding, mental health and food insecurity. Student groups then created websites focused on a particular issue. For example, one group focused on the opioid overdose issue, creating a website that streamlines Narcan vending machine management in Dearborn by enabling QR-based status reporting, centralized monitoring for public health staff, and user-friendly access to machine locations and resources. Another group of students created FloodSafeDearborn, which provides city residents with real-time flood updates, historical data and reporting tools via an interactive map, personalized alerts and statistical insights. Another group created a skillsharing website, where residents can connect with repair resources, local businesses and skill-sharing forums, reducing waste and promoting repairing of consumer goods. Song then invited Hachem back at the end of the semester to help grade the projects.
Both Song and Hachem give the collaboration high marks. Song surveyed students and found that they were more engaged when their learning was connected to a real-world issue so close to home. And he especially appreciated that Hachem went the extra mile and continued to interact with students throughout the semester when they had follow-up questions. Hachem says she appreciated the chance to work with students on creative technology solutions for practical problems. “I love new innovative technology. So collaborations like this are exactly what I want to be doing more of,” she says. “Being able to take specialized skills from everyone — and inspire students too — you have to love that.” She says it was also fun to be on the other side of the class, giving a lecture, when she was in the students’ shoes just a couple years ago.
Song and Hachem recently authored a paper about their collaboration, and he now has plans to build out a similar practice-based learning course where, like SpeedEye, students would build technologies that address specific community needs. He says the turn his career has taken, where he’s putting more energy into community-based collaborations, isn’t something he necessarily expected, but he’s really enjoying it. “Almost 10 years ago, I was working on sensing, and to me, that was just technology,” he says. “But, now, I really like this idea of working with others, like Hayat, and I’m always thinking about planning more activities and projects like this. Connecting technology to the world, that’s how we make an impact. We’re helping the students, helping the environment, helping residents to use this knowledge that we have. And for me, that’s something I can feel good about too.”
###
Story by Lou Blouin. Photo by Matthew Stephens
