From fire-detecting drone swarms to optimally efficient human-autonomy collaboration, the UC Davis mechanical and aerospace engineer uses complex technological systems to address complex challenges.
Smoke is the age-old indicator of a wildfire in California. When caught by camera, satellite or watchtower, visible smoke gives authorities the signal to evacuate affected populations and send out containment forces. But in the time it takes to spot the smoke, the flames may have the chance to blaze across acres of land.
Now, imagine: A network of chemical sensors detects a small seeding fire and triggers the deployment of remote-controlled scouting drones. An expert at central command reviews the data to confirm a fire, and authorities receive an alert within minutes, rather than hours. Firefighters reach the scene more quickly, and the blaze is contained before the smoke becomes visible.
This is one of the technological solutions for tangible, real-world issues envisioned by Zhaodan Kong, an associate professor of mechanical and aerospace engineering at the University of California, Davis, and a principal investigator (PI) at the Center for Information Technology Research in the Interest of Society and the Banatao Institute (CITRIS). His UC Davis Cyber-Human-Physical Systems (CHPS) Lab explores cyber-human-physical systems — the complicated ways in which people and computers interact via physical devices — and examines the use of artificial intelligence (AI) to pursue more complex goals.
As the role of technology in our daily lives evolves, safe and successful partnerships between humans and machines grow increasingly critical to meeting the scale of society’s needs. Kong works to develop systems with autonomy, or the ability to function independently in uncontrolled environments. His ultimate goal is technological autonomy that is both assured, meaning its behavior is mathematically predictable, and collaborative, meaning it can respond to input from humans, as well as other machines.
His group investigates a variety of ways in which to improve and implement assured and collaborative autonomy, including human-autonomy teaming, which pairs people with AI-enabled systems in pursuit of a common goal; neural engineering, which integrates biomedical science with engineering principles to understand and improve the function of the human nervous system; and the control and application of uncrewed aerial vehicles (UAVs), aka drones.
“I believe AI and other technology can be tremendously helpful to solve problems that we are going to face in the future,” Kong said, “but we have to be careful about the implications. In some situations, we need a human in or on the loop to make sure that the implementation is ethical, safe, and secure.”
Kong earned a bachelor’s and master’s degree in astronautics and mechanics from the Harbin Institute of Technology in northeastern China. After completing a doctorate in aerospace engineering and mechanics at the University of Minnesota Twin Cities and then a postdoctoral position at Boston University, he joined the UC Davis faculty in 2015.
There, he works alongside fellow CITRIS PI and aerospace and mechanical engineer Stephen Robinson at the UC Davis Center for Spaceflight Research and the NASA-sponsored Habitats Optimized for Mission of Space Exploration (HOME) Space Technology Research Institute. He is also affiliated with the NSF-funded AI Institute for Next Generation Food Systems, the UC Davis Air Quality Research Center, the UC Davis Center for Neuroengineering & Medicine, and the UC Davis Wildfires Research Working Group.
Catching wildfires before they spread
California’s unique climate — a perfect storm of wet winters, hot dry summers and frequent drought — has ensured a constant cycle of wildfires over the decades. The severity of the fires, however, has escalated in the 21st century, as climate change has pushed temperatures to record heights. In 2020, the August Complex became the largest fire in the state’s recent history, devastating over 1 million acres across seven counties in Northern California and contributing to a total of more than 4 million acres burned in a record-setting fire season. Economic losses for the year totaled at least $19 billion. The thousands of wildfires that occur every year both result from and perpetuate the worsening effects of climate change, intensifying the urgent need for effective control measures.
Zhaodan Kong’s first focus on the topic of wildfires was addressing the rising health risks they posed. His lab’s initial goal was to develop a swarm of UAVs to sample chemicals in wildfire smoke at the wildland-urban interface (WUI), where smoke is more toxic due to burning structures and vehicles with synthetic components. Ever solution-oriented, however, Kong was simultaneously thinking of ways to address the danger at its source.
“My background is engineering, and we like to solve problems,” he said. “So, you have this destructive wildfire: Is there a way to somehow avoid it?”
Kong and his team turned their attention to deploying the drone swarm as an early wildfire detection and alert system.
Current wildfire detection methods, while extensive, are largely reactive. They also suffer from limits in coverage, as watchtowers and cameras have a finite range of sight, and poor visualization, as satellites can discern fires but not pinpoint locations. Drones, on the other hand, are uniquely maneuverable and easily made autonomous.
With support from a 2021 CITRIS Seed Award, Kong has built an eight-blade rotorcraft, or octocopter, to serve as the swarm’s primary fire detector. UC Davis Health exposure scientist Deborah Bennett and UC Davis engineers Anthony Wexler and Thomas Young are co-investigators on the project.
Their proposed process starts with a ground-based network of sensors originally developed by Wexler, placed in regions of high fire risk. In the researchers’ ideal scenario, the sensors pick up the chemical signature of wildfire smoke, and a team of drones automatically deploys to the site, equipped with sensors searching for elevated rates of particulate matter and carbon dioxide. Drone-mounted cameras capturing both visible light and thermal signatures, help to visualize areas that may be obscured by smoke. The swarm’s real-time data goes to the California Department of Forestry and Fire Protection (CalFire), and if a burn is confirmed, CalFire sends out containment forces immediately.
“If you can reduce the response time by 15 minutes or half an hour, you can save lots of lives, property and even money for the state,” said Kong.
As Prabhash Ragbir, a doctoral student in the CHPS Lab, described, a data-driven model lies at the heart of this procedure, creating a chemical, temperature and wind map that forms the basis for the onboard control system, alongside a tracking algorithm that guides the drone to the source of a fire.
To take the project from proof-of-concept to implementation-ready, Kong and crew are collaborating with UC Davis aerodynamicist Camli Badrya to design a tilt-rotor drone that will double flight time from 30 minutes to one hour, enabling the craft to collect more data and reach more remote areas.
Measuring trust in tech
Another potentially lifesaving avenue of investigation for Kong is measuring human trust in technology, especially in systems where safety is on the line. If a person is flying an aircraft or operating a nuclear power plant, misunderstanding the technology’s capabilities, even briefly, could lead to fatal accidents.
To help human-autonomy teams collaborate more effectively, Kong is working with UC Berkeley mechanical engineer Kosa Goucher-Lambert to better quantify, and eventually predict, user trust with support from a 2022 CITRIS Seed Award. In their experiment, a human interacts with an autonomous system to perform simple tasks while wearing nonintrusive sensors that track physical attributes such as brain activity, gaze, pulse and heartbeat. This allows the researchers to collect feedback from the human subject without disrupting them from the flow of the task. They are also asked periodically about their trust in the system.
By adding a layer of physiological metrics atop the gold-standard survey questions, the team aims to calibrate the user’s trust in real time, with the goal of designing systems that strike an optimal balance between overreliance and skepticism. For instance, if a person exhibits low trust, the system can offer an explanation that increasings their understanding of the technology’s abilities and, as a result, their level of assurance.
“The reason we’re focusing on trust right now,” said Kong, “is that it’s a very hard thing to infer.”
If his team can successfully predict human-machine trust, they plan to expand their model to further infer how users are affected by varying workloads and situational awareness. Ultimately, they intend to use that information to design more efficient and effective human-autonomy teams.
Their project, now in its second year, has grown from its CITRIS-seeded roots to include researchers from the University of Colorado Boulder, with further support from the U.S. Air Force Office of Scientific Research (AFOSR) and NASA.
Supporting emerging leaders
As he works to develop collaborative relationships between humanity and machines, Zhaodan Kong also takes care to support the next generation of researchers, for whom human-technology interactions will be even more prevalent.
In summer 2024, Kong’s lab participated in the CITRIS Workforce Innovation Program, where host organizations, such as academic laboratories and tech startups, train students to build skills in leadership, project management and areas of emerging innovation.
Luuanne Chau, an undergraduate studying computer science at UC Davis, “dipped her toes,” as she described it, into the realm of machine learning as an intern on the wildfire project, where she developed a model to detect a wildfire in its early stages using thermal imaging.
“Although Dr. Kong was busy, he set aside time for me to meet with him every week, where I could discuss what I’m trying, what’s working and what’s not working, and then he could give me suggestions,” she said.
As someone who had never previously considered working with drones, Chau recalls her time with Kong and his lab group as an affirming experience where she felt she contributed to something meaningful.
“He’s an understanding PI who cares for his students and has given me a lot of freedom with this project,” she said. “He’s also presented some opportunities for me to pursue for the upcoming school year, which I really appreciate.”
Looking forward, looking to the skies
Kong intends to keep pursuing collaborative solutions to complex challenges, with CITRIS Aviation as a promising new venue. He was just named to the research initiative’s faculty and staff leadership committee.
He cites CITRIS’s multicampus presence and interdisciplinary mission as a multiplying strength.
“There’s no way that I could solve these problems by myself,” Kong said. “But by combining different groups, expertise and experience — we all have to work together to create solutions that are not only effective but also safe, ethical, sustainable and beneficial for everyone.”