In 2022, Randall Pietersen, a civil engineer in the U.S. Air Force, set out on a training mission to assess damage at an airfield runway, practicing “base recovery” protocol after a simulated attack. For hours, his team walked over the area in chemical protection gear, radioing in geocoordinates as they documented damage and looked for threats like unexploded munitions.
The work is standard for all Air Force engineers before they deploy, but it held special significance for Pietersen, who has spent the last five years developing faster, safer approaches for assessing airfields as a master’s student and now a PhD candidate and MathWorks Fellow at MIT. For Pietersen, the time-intensive, painstaking, and potentially dangerous work underscored the potential for his research to enable remote airfield assessments.
“That experience was really eye-opening,” Pietersen says. “We’ve been told for almost a decade that a new, drone-based system is in the works, but it is still limited by an inability to identify unexploded ordnances; from the air, they look too much like rocks or debris. Even ultra-high-resolution cameras just don’t perform well enough. Rapid and remote airfield assessment is not the standard practice yet. We’re still only prepared to do this on foot, and that’s where my research comes in.”
Pietersen’s goal is to create drone-based automated systems for assessing airfield damage and detecting unexploded munitions. This has taken him down a number of research paths, from deep learning to small uncrewed aerial systems to “hyperspectral” imaging, which captures passive electromagnetic radiation across a broad spectrum of wavelengths. Hyperspectral imaging is getting cheaper, faster, and more durable, which could make Pietersen’s research increasingly useful in a range of applications including agriculture, emergency response, mining, and building assessments.
Finding computer science and community
Growing up in a suburb of Sacramento, California, Pietersen gravitated toward math and physics in school. But he was also a cross country athlete and an Eagle Scout, and he wanted a way to put his interests together.
“I liked the multifaceted challenge the Air Force Academy presented,” Pietersen says. “My family doesn’t have a history of serving, but the recruiters talked about the holistic education, where academics were one part, but so was athletic fitness and leadership. That well-rounded approach to the college experience appealed to me.”
Pietersen majored in civil engineering as an undergrad at the Air Force Academy, where he first began learning how to conduct academic research. This required him to learn a little bit of computer programming.
“In my senior year, the Air Force research labs had some pavement-related projects that fell into my scope as a civil engineer,” Pietersen recalls. “While my domain knowledge helped define the initial problems, it was very clear that developing the right solutions would require a deeper understanding of computer vision and remote sensing.”
The projects, which dealt with airfield pavement assessments and threat detection, also led Pietersen to start using hyperspectral imaging and machine learning, which he built on when he came to MIT to pursue his master’s and PhD in 2020.
“MIT was a clear choice for my research because the school has such a strong history of research partnerships and multidisciplinary thinking that helps you solve these unconventional problems,” Pietersen says. “There’s no better place in the world than MIT for cutting-edge work like this.”
By the time Pietersen got to MIT, he’d also embraced extreme sports like ultra-marathons, skydiving, and rock climbing. Some of that stemmed from his participation in infantry skills competitions as an undergrad. The multiday competitions are military-focused races in which teams from around the world traverse mountains and perform graded activities like tactical combat casualty care, orienteering, and marksmanship.
“The crowd I ran with in college was really into that stuff, so it was sort of a natural consequence of relationship-building,” Pietersen says. “These events would run you around for 48 or 72 hours, sometimes with some sleep mixed in, and you get to compete with your buddies and have a good time.”
Since coming to MIT with his wife and two children, Pietersen has embraced the local running community and even worked as an indoor skydiving instructor in New Hampshire, though he admits the East Coast winters have been tough for him and his family to adjust to.
Pietersen went remote between 2022 to 2024, but he wasn’t doing his research from the comfort of a home office. The training that showed him the reality of airfield assessments took place in Florida, and then he was deployed to Saudi Arabia. He happened to write one of his PhD journal publications from a tent in the desert.
Now back at MIT and nearing the completion of his doctorate this spring, Pietersen is thankful for all the people who have supported him in throughout his journey.
“It has been fun exploring all sorts of different engineering disciplines, trying to figure things out with the help of all the mentors at MIT and the resources available to work on these really niche problems,” Pietersen says.
Research with a purpose
In the summer of 2020, Pietersen did an internship with the HALO Trust, a humanitarian organization working to clear landmines and other explosives from areas impacted by war. The experience demonstrated another powerful application for his work at MIT.
“We have post-conflict regions around the world where kids are trying to play and there are landmines and unexploded ordnances in their backyards,” Pietersen says. “Ukraine is a good example of this in the news today. There are always remnants of war left behind. Right now, people have to go into these potentially dangerous areas and clear them, but new remote-sensing techniques could speed that process up and make it far safer.”
Although Pietersen’s master’s work primarily revolved around assessing normal wear and tear of pavement structures, his PhD has focused on ways to detect unexploded ordnances and more severe damage.
“If the runway is attacked, there would be bombs and craters all over it,” Pietersen says. “This makes for a challenging environment to assess. Different types of sensors extract different kinds of information and each has its pros and cons. There is still a lot of work to be done on both the hardware and software side of things, but so far, hyperspectral data appears to be a promising discriminator for deep learning object detectors.”
After graduation, Pietersen will be stationed in Guam, where Air Force engineers regularly perform the same airfield assessment simulations he participated in in Florida. He hopes someday soon, those assessments will be done not by humans in protective gear, but by drones.
“Right now, we rely on visible lines of site,” Pietersen says. “If we can move to spectral imaging and deep-learning solutions, we can finally conduct remote assessments that make everyone safer.”
Source
