Building A More Resilient City of Angels, Part 2

USC Viterbi Staff | February 3, 2025 

From fire response robots to retardant firing artillery, USC Viterbi engineers offer potential solutions for the future.

Firefighting Robot

Editorial animation by Runway ML.

In early January, Governor Gavin Newsom said: “November, December, now January − there’s no fire season, it’s fire year. It’s year-round.”

Facing the uncertainties of droughts, wildfires, blackouts, poor air quality, mudslides and flooding, we canvassed the faculty of the USC Viterbi School of Engineering for possible solutions.

Here are some ideas to build a more resilient City of Angels (read the previous story here):

“Rethink Before We Rebuild”

Lucio Soibelman
Fred Champion Estate Chair in Engineering Professor

Lucio Soibelman

Lucio Soibelman

“We should see this as an opportunity to rethink our urban environments before we rush to rebuild everything the same way that it was before the fires.

First, we need to make sure that the hills near the burned areas are stable. It is expected that those mountains, now with burned vegetation cover, won’t be stable during future rains. The consequence will be the expected landslides that will affect the same neighborhoods that were destroyed by the fires. It would be a disaster to rebuild just to have the new houses destroyed again by landslides.

We have to rethink the interface of neighborhoods with wild areas. In the last 20 years, we built too many houses on those areas, increasing the danger of fire disasters. We are not prepared nor equipped to deal with the increased danger of this interaction. I think that it’s time to stop new construction in those areas.

When we rebuild the burned neighborhoods, we will need to rethink the way that we design and build our landscapes. There is a need to make sure that we select better plants with different burning characteristics. Palms burn much faster than oak trees. We need to select vegetation that doesn’t burn as fast as the ones that we have today. The same is the need to keep vegetation a good distance from our homes with paved areas separating our garden from the house. We must be more careful with fire bridges like wooden fences, gates or trash near homes.

We will need new building codes to take this in consideration. From the landscape to the selection of building materials. We need to design houses with no combustible sidings or with building materials like CMU (precast, rectangular concrete) blocks or concrete walls. I think that concrete 3D printing could be a solution that will be evaluated for building more fireproof houses. Same for roofs: we need to avoid materials like asphalt shingles (that can be manufactured to resist fire, but even those are combustible). A better solution would be to use metallic roofs. All openings will need to be upgraded from fireproof doors to high quality, multiple pane windows. But research is needed to understand the real cost benefit of designing new fireproof houses and retrofitting old ones to resist fires. Even with all these upgrades, there is the is no guarantee that a house will not burn. Many times, houses and businesses burn from the inside out. If you have any small openings like vents, ambers flying on strong Santa Ana winds will find a way into the house and thousands of combustible materials to ignite a fire from the inside out. There is a need to build houses that have all openings covered with special meshes that would disallow even the smallest embers.

But even if we have a very fireproof house, it will probably burn if all the houses around it are burning, making the temperature so high that some indoor materials could combust naturally! So, improving resilience to fire is a community task and not just an individual one. If you have a house with fireproof materials, but your neighbor abandons his car with a tank full of gas on his driveway next to your home — when his house burns and his car explodes, the fireproofing of your house will probably not be enough.

Another lesson learned is to rethink the transportation network serving these communities. During the fire, we could see that the escape routes were not designed for the demand needed during an emergency. Many people had to abandon their cars when they were trying to escape the fires because they got stuck on traffic. Several areas of the city that are prone to fire like the Hollywood Hills have very narrow streets that are extremely dangerous to be used as escape routes during fast fires.”


“Fighting Wildfires with Early Detection and Artillery”

Peter Beerel
Professor of Electrical and Computer Engineering

Peter Beerel

Peter Beerel

Mitul Luhar
Henry Salvatori Early Career Chair and Associate Professor of Aerospace and Mechanical Engineering and Civil and Environmental Engineering

Mitul-Luhar-portrait

Mitul Luhar

 

 

 

 

 

 

 

 

Barath Raghavan
Associate Professor of Computer Science and Electrical and Computer Engineering

Barath Raghavan

Barath Raghavan

“LA-96C was a Nike missile station located off Mulholland Drive. Now part of San Vicente Mountain Park and a popular hiking destination, in the 1950s, it was part of a network of radar installations designed to detect incoming enemy

bombers. It used millimeter wave communication to connect with a missile base located in the Sepulveda Basin that could launch missiles designed to destroy enemy bombers before they could drop their destructive payloads. We believe this combination of early detection and rapid response provides a roadmap for a radical approach to fight wildfires.

Whereas there are numerous efforts to use a network of cameras and watchtowers, some manned and others with semi-automated AI-assisted technology, their coverage is far from perfect, enabling wildfires to get out of control quickly. These systems need to be enhanced with more powerful, fixed-wing aerial drones that can tolerant high winds and stay

aloft for hours, community-hosted cameras that are networked and enable cheap improved coverage, as well as new tethered, drone-in-the-box systems that can tolerant the high Santa Ana winds. These systems need to provide close to 100% coverage, so they detect wildfires as they emerge and provide a record of any arsonist responsible for their start.

Whereas early detection is within our reach, the time for fire fighters to get to the location is often so large that the fire is already out of control, particularly in the rugged slopes of the Southern California foothills. One potential but radical solution is fire-retardant artillery. In particular, shells can be filled with retardant and directed at the fire with pin-point accuracy using altitude-aware fuses that ensure effective airborne delivery of the retardant. Just like the Ajax missiles of the 1950s, these shells can be stationed miles away from the mountain in locations such as in the Sepulveda Basin. Another potential advantage of fire-retardant shells is

that they could enable effective fire suppression in dangerous high-wind conditions that ground the aerial firefighting fleet.

Moreover, with AI-powered cameras that can track embers, these retardant delivery mechanisms can be used to not only attack wildfires as they emerge but also target hot-spots caused by embers that can travel miles from the edge of the fire.

There is no doubt such ideas will trigger concerns regarding privacy and collateral damage and raise issues of cost. However, these concerns can be addressed with careful engineering. Coupled with the hardening of homes, large setbacks between houses, fire-resistant vegetative fire breaks, water piped along fire roads, and more aggressive brush clearance, such radical ideas deserve our attention as the current strategies have proven so dramatically to be insufficient.”


“Let’s Optimize: Prescribed Burns, Pre-Positioned Equipment and Smarter Vegetation”

USC computer scientist and CAIS co-director Bistra Dilkina (Photo/Caitlin Dawson)

Bistra Dilkina

Bistra Dilkina
Dr. Allen and Charlotte Ginsburg Early Career Chair in Computer Science and Associate Professor of Computer Science

“To help Los Angeles prepare for the next major wildfire or natural disaster, advanced technology like optimization models and machine learning can improve the city’s resilience. These tools are especially useful when resources are limited, helping decision-makers prioritize the most critical areas to protect, like homes, infrastructure, and natural habitats.

For example, optimization models can help fire managers decide where to conduct prescribed burns to reduce wildfire risks in key areas, such as watersheds and urban zones. My team has developed a tool that combines wildfire simulations with interactive visualizations, allowing decision-makers to explore options and make informed, data-driven choices.

Machine learning takes things a step further by improving predictions of where fires might occur and how they’ll spread. This helps us allocate resources proactively—like positioning firefighters and equipment in areas with high risk—before fires even begin.

Sustainable urban planning also plays a big role in reducing vulnerability. By integrating wildfire mitigation into decisions about where to build and how to manage vegetation, we can make the city more fire-resistant from the ground up.

Advances in AI and remote sensing also help by detecting fires early, with deep learning analyzing data from satellites and drones. These technologies allow us to respond quickly and adjust our plans as conditions change.

Ultimately, collaboration between researchers, government, and local communities is key to making these solutions work. By combining technology, smart planning, and teamwork, Los Angeles can better handle the threat of wildfires and other natural disasters.”


“Public Health Strategies to Stop Catastrophic Fire”

Randolph Hall
Dean’s Professor of Industrial and Systems Engineering, director of Center for Risk and Economic Analysis of Threats and Emergencies (CREATE)

Randolph Hall, Director and Senior Research Fellow at the USC Center for Risk and Economic Analysis of Threats and Emergencies (CREATE)

Randolph Hall

“Wildfires can behave like diseases that are transmitted from person to person, such as Covid-19.

When we protect ourselves against the hazard, we also lower the odds of passing the threat on to our susceptible neighbors. But what should we do? Many structures in Los Angeles were built in ways that expose flammable materials to flying embers, thus elevating their risk of catching fire. We know from decades of public health research that we can reduce and even eradicate disease by increasing rates of vaccination in populations. Likewise, we can be safer if we incentivize ‘vaccination’ of structures against flying embers in vulnerable locations — meaning we should subsidize retrofitting so that structures are less likely to burn and spread fire.

Wildfires, like transmissible diseases, also spread at exponentially increasing rates over time, meaning it is hugely important to stop or contain a fire soon after it is ignited, at its origin. To do so, we need research and innovation on technology that can quickly put out fires in remote locations. This will require surveillance systems to rapidly and accurately detect small fires as well as robotic technology that can extinguish and contain fires where they start — overcoming challenging terrain, dispersion of hazards and extreme winds. Fires need to be stopped in the precious minutes before they become catastrophic.

History has shown that public health investment in vaccination and disease surveillance can save lives. We now need a public health strategy for wildfires, stopping their spread and saving communities by ‘vaccinating’ structures against burning embers and rapidly detecting and extinguishing fires at their origin.”


“Robots For Wildfire Surveillance and Response”

Quan Nguyen

Assistant Professor of Aerospace and Mechanical Engineering and Computer Science

Quan Nguyen

“My research focuses on legged and humanoid robots, and we really hope that in the near future our robots can help with natural disasters in terms of surveillance, prediction and disaster response.

Legged and humanoid robots have the capability to overcome extreme terrain that’s normally unavoidable during natural disasters. They can help go around remote forests tirelessly to check the forest conditions and help predict and prevent the risk of wildfires. When the fire happens, they can help to carry fire distinguishes to help reduce the fire. However, it will require coordination of a large team of robots and in collaboration with human firefighters to do it effectively.”


“Fighting Social Media Misinformation During Crises”

Kristina Lerman
Research Professor of Computer Science and Principal Scientist at USC Information Sciences Institute

Kristina Lerman (Photo/Jeffrey Fiterman)

Kristina Lerman

“During disasters, social media’s role as a primary information source creates significant challenges for crisis communication and emergency response.

While enabling rapid information sharing, these platforms also facilitate the spread of false claims that can overwhelm official channels and impede response efforts. Our DARPA-funded research addresses these challenges through an innovative solution combining two key technologies: an automated claim detection system using Large Language Models to create real-time knowledge graphs of interrelated information, and a network-based analysis framework leveraging fraud detection algorithms to identify potentially false claims.

We propose to build on this technology to create a dashboard to help journalists and civic leaders efficiently monitor and manage crisis-related claims, reducing the burden of manual fact-checking while strengthening emergency response capabilities.”


“Sensors, Sensors, Everywhere”

Chukwuebuka C. Nweke
Assistant Professor of Civil and Environmental Engineering

headshot of CEE faculty nweke

Chukwuebuka Nweke

“One tangible solution to help Los Angeles build resilience against future wildfires and natural disasters is the development and deployment of a real-time hazard monitoring and predictive analytics network.

This system would integrate advanced sensor technologies to monitor soil stability and other environmental conditions across the city, particularly in high-risk zones like slopes and wildfire corridors. These sensors can include thermal/moisture cameras, LiDAR (Light Detection and Ranging) and others to be permanently deployed spatially for regional coverage. The resulting data could provide the ability to establish predictive models, which, coupled with the network, could provide early warnings, identify vulnerable infrastructure and guide proactive mitigation efforts — such as targeted vegetation management, improved firebreaks and enhanced emergency response coordination.

Ultimately, this approach embodies the principle that ‘you have to be aware to be prepared,’ enabling Los Angeles to anticipate risks and act decisively to safeguard its future. In summary, the key to rebuilding resilience lies in understanding how disasters unfold and using that knowledge to anticipate and mitigate risks. Natural hazards research may provide the tools (sensors) and therefore the data needed to develop predictive models and real-time monitoring systems, enabling communities to rebuild smarter and better prepared for future threats.”


Flying past concrete homes

Editorial animation by Runway ML.

“Codifying 3D Printing Standards for Houses”

Bora Gencturk
Professor of Civil & Environmental Engineering

Professor Bora Gencturk

Bora Gencturk

“When it comes to wildfires, adopting more fire-resistant construction materials must be a community decision to ensure infrastructure that can withstand future events. A single house or group of houses that are fire-resistant will not be enough to achieve this goal.

Concrete is the most widely used construction material and one of the most fire-resistant. Concrete is non-combustible, does not generate smog and helps slow the spread of fire. However, rebuilding entire communities quickly using traditional concrete construction methods presents a significant logistical challenge.

One promising technology in this area is automated robotic construction, also known as construction-scale 3D printing. The Structures and Materials Research Laboratory (SMRL), which I lead, houses a full-scale 3D concrete printer capable of printing single-story structures. SMRL is also the first lab in the world accredited to test 3D-printed construction. Over the years, we have gained extensive knowledge in this technology through testing 3D printing materials and construction methods for various leading companies.

Currently, I am chairing a consensus committee working on developing the first comprehensive global standard for automated 3D construction. This standard is in its first public commenting period, and we are also working to have it adopted into the International Building Code (IBC) during the next revision cycle. Inclusion in the IBC will allow designers and builders to use this technology without needing special permits.

This technology holds strong potential for rapid, durable construction. USC, as a major knowledge hub in Southern California, will undoubtedly play a critical role in the recovery efforts in the months and years ahead.”


“Wildfire Surveillance: A New Public Utility”

Andrew Rittenbach
Computer Scientist, Lead Scientist, USC Information Sciences Institute (ISI)
and
JP Walters
Research Director and Distinguished Principal Scientist, USC Information Sciences Institute (ISI)

Andrew Rittenbach

Andrew Rittenbach

Image

JP Walters

“Wildfires pose a growing and enduring humanitarian threat, both worldwide and in Los Angeles. They can start anywhere but are particularly destructive at the wildland urban interface (WUI), a common feature in many parts of L.A. As a result, we must identify the wildfires in as close to real-time as possible, wherever they are.

However, early detection of fires at the WUI has been a particularly challenging problem. Often, structures like reflective rooftops or solar panels are incorrectly identified as fires when using physics-based fire detection algorithms, which apply image thresholding using human-designed criteria.

In our work, we’ve demonstrated novel computer vision-based AI algorithms that accurately detect fires while reducing the number of false alarms at the WUI. This early detection is vital but can only be accomplished through ongoing wildfire surveillance, requiring a coordinated aerial imagery system composed of satellites, drones and more. Satellites can provide persistent coverage and rapid revisit rates across a heterogeneous set of sensors.

Our results have demonstrated accurate fire detection across sensors and modalities. After detecting the wildfire, drones may provide high-resolution imagery capable of detecting embers and more while providing responders with up-to-the-minute imagery. This system requires real-time coordination, analysis and algorithm development, which is now within reach.”


“Modifying Weather Conditions”

Geoffrey Spedding
Professor of Aerospace and Mechanical Engineering

Geoff Spedding

“One interesting aspect of the recent fire outbreaks is that the Santa Ana winds themselves are taken as given, but it is natural to wonder whether these strong driving events themselves could be modified. Attempting to modify weather conditions themselves may be seen as a foolish or hopeless quest, but it could be interesting to think about the genesis of the disruptive winds.

The first condition is a set of temperature variations far off in the flatlands of Nevada and Utah, where the event is initiated.

Local climate modelers know this well and predictions of wind strength seem to be mostly correct. I wonder what kind of force would have to be applied, over what kind of area, to significantly change these generating conditions.

Also, in the Palisades fire, particularly destructive wind forces seem to have come when the downslope accelerations generated a large degree of variability and turbulent motion, famously sending burning embers upwards and then sideways over long distances.

In an armchair engineering world, one would like to be able to put devices into the flow at the crests of the mountain ranges. The goal would be to deliberately generate turbulence to reduce the likelihood of the creation of larger-scale turbulence downstream that lifts embers and other things into the air and transports them laterally. This is a ubiquitous feature in wind tunnel design, where upstream wire mesh screens produce small scale, dissipative turbulent fluctuations that quickly die down to leave a smooth and turbulence-free flow downstream.

Even if putting up similar screens were possible, no-one would tolerate the spoiling of archetype views of mountain ranges. Still, I wonder what kind of deliberate disturbance might be created by a fleet of flying devices trailing obstacles that create initial turbulence?”


“Designing Homes in the Era of Mega-Fires”

David Gerber
Professor of Practice in Civil and Environmental Engineering and Architecture

David Gerber

David Gerber

“There are four core areas I’m involved in: design, innovation, technology and entrepreneurship. Design matters: at a minimum, there are four primary things we can address to protect buildings against fire. Class A roofing systems (those that have the highest level of fire resistance as classified by building codes and testing standards). Fire retardant siding (exterior cladding) and building envelopes as the first line of defense. Insulated glass windows. Defensible space designed to slow the spread of fire, protect the structure, and provide firefighters with a safer area to contain a blaze.

When it comes to infrastructure, we can significantly reduce risks by burying our electrical transmission lines, updating our water and distribution systems and taking strategic measures within city and regional planning.

Supporting innovation and entrepreneurship will also be crucial. This is a chance to take advantage of novel building materials and apply the latest tools for simulation and risk illustration. Large scale destruction calls for a more efficient approach to building back homes and schools, and novel modular and industrialized construction solutions can significantly compress the time and cost of reconstruction.”


“War Games for Firefighting Aviators”

Tom Anthony
Director, USC Aviation Safety and Security Program

Thomas Anthony, director of the USC Viterbi Aviation Safety and Security Program (Photo/Courtesy of Thomas Anthony)

Thomas Anthony

“Currently, the county has little interaction between sheriff and fire aviation. So, the city and county could each create an Office of Aviation Operations Manager.

While fire and rescue aviation units will hold joint exercises for search and rescue missions, they do not hold joint exercises for firefighting missions; at least not one that either I or the firefighter aviators I have spoken to know of.

A large event — perhaps similar to a military training exercise — could bring multiple aerial firefighting units to Southern California to scoop water and practice putting out mock fires. This would test plans for a multi-agency response without waiting for a large fire.”

 


“Finding the Best Routes for Evacuees and First Responders”

Cyrus Shahabi
Helen N. and Emmett H. Jones Professorship in Engineering and Professor of Computer Science, Electrical and Computer Engineering, and Spatial Sciences
and
John Krumm
Associate Director, Integrated Media Systems Center (IMSC)

Cyrus Shahabi

Cyrus Shahabi

John Krumm

John Krumm

“For localized natural disasters, it is critical to get affected victims out and emergency responders in. Roads are stressed by both the disaster and the evacuees, degrading the accessibility of the areas that need it most.

In our lab, we have developed deep expertise in understanding how people move around based on terabytes of cell phone GPS data, detailed maps of ground features, and advanced AI techniques for modeling human mobility behavior. Working with wildfire experts, we propose to carefully model likely fire scenarios and evacuation behavior to anticipate how they affect the accessibility of routes into and away from the fires. These models will be used in multiple ways to reduce the harm of wildfires:

We can identify neighborhoods that are most likely to become trapped by wildfires, based on their fire risk and likelihood of blocked access. This can inform where to invest in fire prevention and evacuation assistance.

Firefighters are already pre-deployed in anticipation of fires. Our models can optimize this practice by finding the best use of resources in light of likely fire locations, possible route blockages, and backup plans if the primary responders cannot access the fire location.

Use GPS data from evacuees, we can monitor evacuation routes in real time for optimal rerouting. We can predict which routes will become blocked and route around them.

Evacuees without access to a personal vehicle may not be able to drive themselves away from the fire. We can coordinate ride sharing among neighbors as well as public transit for emergency evacuations of people who cannot otherwise get away.”

Published on February 3rd, 2025

Last updated on February 3rd, 2025

Share this Post