Yun Huang's MapInsights helps police trainees learn the territory

In early 2026, the University of Illinois Division of Public Safety was facing a challenge. They had 11 security guards they planned to convert to community service officers, which meant they would change from being based in one building to working a whole beat. 

"Their jobs completely changed," said Lieutenant Ty Ledbetter, administrative lieutenant for the division, who also manages the community service officer program and officer training. Public Safety needed a way to get the officers up to speed, and quickly. Their solution was to call Associate Professor Yun Huang, who ultimately built the MapInsights program.

Huang, whose research focuses on human-AI collaboration and interactions, was already at work on interactive mapping software. She started working with Ledbetter and Jennifer Whiting from the university's Police Training Institute in March, and by May had created MapInsights, a web-based training platform that converts the memorization of a large real-world operating environment into structured, measurable practice.

For a public safety officer, knowing every building on a beat is essential. However, memorizing more than a thousand buildings on the University of Illinois campus, which is divided into three beats on roughly a north-to-south basis, can be tedious and time-consuming. Furthermore, the division's test of campus geography, on which trainees must score 80 percent or higher before moving to the next training module, was a clunky webform that didn't even save progress.

In contrast, MapInsights is highly reactive. MapInsights identifies each building by name but also by the number used by the campus Facilities & Services office. All maps are connected to the university's archaeological services, so if a building number changes in the system, it will change on the map. When the assessment begins, all buildings are outlined in red. Each time a user clicks on a building and identifies it correctly, the red changes to green. Progress is savable, and the interface is reactive. The design turns campus geography into a short practice loop: each round shrinks the map to one beat and either buildings or parking lots, shows a highlighted target, and offers a short list of names, then brings back only the misses until the officer can recognize those places with less help.

The system design was informed by cognitive psychology research. For example, it incorporates a sense-of-direction assessment framework to help measure the trainee's sense of direction, map reading, distance judgment, route memory, orientation in novel places, and mental map formation. With that information, the training can adapt to each user's abilities and tendencies.

"This generation," Ledbetter said as he clicked through a practice exam, "this is exactly what they're used to doing."

MapInsights has become part of training for both sworn police officers and community officers in Public Safety. With some additional mapping, Ledbetter envisions possible training for campus SafeWalks escorts and even publicly available maps showing the best-lit or most-accessible walking routes across campus.

"It's really affected the department as a whole," he said, calling Huang's work "phenomenal."

Meanwhile, Huang sees other applications of MapInsights. For example, it could allow new students to upload their course schedules, showing them the best and quickest routes to classes—or whether they even have enough time to get from one class to the next based on their schedule. 

Huang has worked to design technology for public safety since 2010, including SafeRBot, a large language model that helps dispatch centers respond to non-emergency situations that was supported by the University of Illinois Proof-of-Conceptprogram. The work on SafeRBot contributed to a patent related to LLM-assisted incident reporting and helped establish lasting collaborations between Huang's research team and local public safety stakeholders.

She potentially sees the two applications working together.

"We want to turn this research platform into something really practical that people can use," she said.

Huang's experience with MapInsights also reflects how AI is reshaping the role of experienced researchers. During her PhD and postdoctoral training, Huang developed extensive experience in hands-on software development and system building, deploying systems in large cities and analyzing years of systems logs. As her academic career progressed, her role naturally shifted toward leading research projects, mentoring students, and guiding system design. With advances in AI-assisted programming, she has found herself returning to hands-on development, bringing together years of technical expertise, research experience, and design vision to build and deploy systems directly.

"It's exciting to be back in the creator's seat," Huang said. "AI allows me to return to hands-on development while doing intense research and advising, combining the technical skills I've built over the years with a much deeper understanding of real-world problems. I can navigate between research and reality smoothly and see immediate impacts."

These partnerships continue to create new opportunities. For example, Huang has collaborated with the Police Training Institute on a proposal to the US Department of Justice to extend MapInsights technology to geographic training for municipal police departments. She has also begun exploring projects with Brian R. Brauer, executive director of Emergency Management at the University of Illinois, including the development of emergency action plans that could leverage the MapInsights platform to enhance preparedness and community safety efforts.