Confronting Wildfire: Into the Fire to Advance Research & Innovation

Wildfires are larger, burn longer, cause more damage to structure, and harm more people than ever before. Today, they create challenges for homeowners and firefighters throughout the state of California and increasingly, throughout the World. Labs, researchers, and entrepreneurs at the University of California are tackling this massive problem through studying how these fires start, grow, and most importantly – how they can be detected early and rapidly respond. For all these groups, access to live wildfire events is not possible due to risk, rules, and regulations. This means it is difficult to validate a hypothesis, test a technology outside of the lab, and get feedback from key stakeholders, such as firefighters, on the group and leadership with decades of experience in responding to fires across the state.

In spring 2021, Dr. Thomas Azwell, director of the Disaster Lab, brokered a unique opportunity for the Combustion Modeling Lab, Berkeley Fire Research Lab along with two UC Berkeley innovations, Gridware and Squishy Robotics, to have access to a significant fire event. Over two hundred acres were set ablaze by Marin County Fire Department for a controlled burn which created a fire break in the wildland-urban interface above the town of Novato.  Not only did the researchers and startups attend the event, but they also collected invaluable data to advance their innovations and received feedback from key stakeholders as to what data was important to them and how their technologies may be utilized during a future fire event.

Gridware’s approach is to use a small, sensor-laden box that can be installed on a power pole with just four screws. Gridware’s package contains microphones and other sensors to listen to the ambient environment around a power pole, and it uses on-board AI/ML processing to detect anomalies and report them to the relevant authorities. “This opportunity allowed us to put our system out in the real world, to not only collect real information but also to see how different subsystems handle being put in a volatile environment. The information we collect is extremely valuable to our fire detector development and we would jump on any future opportunities to collect more data.”, said Patrick Chwalek (Gridware). 

Squishy Robotics, a spin-off of research that came out of Professor Alice Agagino’s BEST Lab at UC Berkeley, is a sensor robot that can be air-deployed into hazardous areas to furnish persistent, ground-level, real-time data for emergency situations such as hazardous substance spills, structure, and wildland fires. Squishy Robotics viewed their participation in the training exercise as invaluable. “We were able to observe our sensor robots perform in a real-life wildfire—that is an opportunity that doesn’t come around every day,” Doug Hutchings remarked. “Our engineering team has completed preliminary analyses that show that our robots can detect early wildfires under the right circumstances, and we are taking a deeper look to further improve our software and hardware.”

Dr. Charles Scudiere of the Combustion lab found that attending the controlled burn helped to bridge the gap between theory and real wildfire conditions. “Attending this event greatly bridged a number of gaps between small-scale experiments frequently performed in laboratory settings with the conditions experienced during a wildfire, and helped identify, focus, and motivate future risk and wildfire work. Observing the influence of the underlying small-scale key research areas influence the macroscale fire burning provides a unique perspective and validation check in analyzing wildfire and WUI data, such as when modeling and using the ELMFIRE wildfire modeling tool built into Pyregence. For instance, the mosaic-shaped burned area of the hillside, in part attributed to the fuel moisture content and wind humidity was different from what is typically targeted in controlled laboratory-scale experiments.”

Dr. Azwell is continuing to create opportunities for fire-related startups and research labs to collect data, get constructive feedback, and prove their concept with key stakeholders and real-world situations. He is actively expanding the research program for the Disaster Lab to include developing, demonstrating, and deploying life-saving technologies that have the potential to mitigate the damage caused by catastrophic environmental events.