During Sunday’s GEOINT Foreword, a diverse panel of subject matter experts spoke about how the combination of artificial intelligence (AI) and geospatial data could aid in pre-disaster evacuations, disaster relief, and more.
Brian Collins, CEO of Intterra, began the conversation by speaking about the need for rapid analysis of the vast quantities of data fed to first responders, specifically firefighters.
“We need to assess the large amounts of data that come in and we need to start looking at it before we hand it off to [first responders],” Collins said. “We need rapid analysis of that data instead of a trigger of alerts.”
He further explained how repurposing this information would allow evacuation pre-planning, rapid predictions regarding where the fire is expected to spread, and assessments of the social and economic impacts.
Eli Ibanga, a student at the University of Southern California’s Spatial Sciences Institute, discussed how AI helps with evacuation preparedness ahead of a disaster.
“We wanted to focus on the preparation as opposed to response and recovery, which we often spend a lot of our time and resources in,” Ibanga said.
Applying AI for evacuation reduces recovery costs, loss of life and injury, and community recovery time. One example Ibanga gave was health information exchange and the digitization of medical records.
“By leveraging machine learning and artificial intelligence the data can be processed and coded to come up with solutions for potential barriers to recovery or [to identify] people who could be at risk,” Ibanga said.
William Porter, senior manager of operations support at nonprofit Team Rubicon, spoke about his team’s response following several disaster scenarios such as Hurricane Florence in September 2018.
Team Rubicon members traveled to North Carolina to collect and map data. The information included flood zones, functioning gas stations, distribution points, open roads, etc.
“It impacted the traditional response and informed decisions,” Porter said.
Jessica Hulsey, a product development manager at BAE Systems, spoke about understanding AI and machine learning in the context of disaster response.
“What we are trying to understand in disaster response and geospatial intelligence is how we can apply these capabilities to improve how we approach problems,” she said. “ … What are the questions we need to ask? What are the answers we are looking to get in order impact response in a meaningful way?”
She showed imagery captured during Hurricane Florence. AI used at the time detected various items on the road, such as fallen trees and standing water.
“But what are we trying to understand in this scene? The human in this can say the tree is not blocking the entrance to the property, the standing water is not approaching the residence, which is important to know,” Hulsey explained. “We are seeing some computer vision techniques but then we want to link that where we can get to decision-making.”
Ultimately, according to Hulsey, it comes down to observation and classification. While AI and geospatial data are complex it’s important to ask the right questions and to identify the data, resources, and techniques that lead to timely answers.