News > Blog

These Berkeley College Students Are Using Artificial Intelligence to Help with Earthquake Response


Don Schaefer, Specialist, Social Media, Consumer Technology Association

College campuses around the world are a key source for engaging and inspiring the next generation of leaders to harness modern, emerging and open source technologies to develop solutions for disaster response and resilience. Companies like IBM are spearheading these efforts by partnering with the Clinton Global Initiative University (CGI U) to provide students with opportunities to use technology for good by co-hosting codeathons at universities primarily in the U.S. as well as in Europe.

The first “Code and Response Codeathon: Presented by IBM and CGI U” was held at UC Berkeley in Spring 2019. We had a chance to engage with the winning team comprised of three French students – Thomas Galeon, Pierre-Louis Missler and Meryll Dindin – who put their ideas into action with a variety of IBM technologies by creating AsTeR: a platform for collecting and prioritizing information to facilitate decision making during natural disasters.

How did you first come up with the idea for your concept? What specific problem were you trying to solve?

Ever since we arrived in California in 2018, people have been telling us about the Big One – the next big earthquake along the San Andreas Fault. If and when it hits, all people living in the area will become victims. As the more than seven million people living in the Bay Area start calling emergency services, call centers will rapidly become overloaded. This makes assigning priority to certain areas unfeasible and prevents emergency responders from focusing their efforts on those most in need. With AsTeR, we aim to create a platform for collecting and prioritizing information in large-scale situations of emergency such as an earthquake or a wildfire.

How does the technology behind your concept work?

To start with, our team focused on the analysis and sorting of emergency calls. When thousands of calls are coming in every minute, call centers will switch to semi-automatic mode: calls will be analyzed in real-time to determine their priority, which will help the first responders to dispatch the rescue teams where they are needed more urgently.

The core technology is based on speech to text conversion as well as information extraction from IBM Watson Natural Language Understanding (NLU). Natural Language Processing (NLP) is used to extract keywords and get relevant information such as address and number of people involved. The information is then added to a visual map updated in real-time to dispatch units in the most efficient way, and the priority of each call is sorted out for a clearer overview. On the other end, we keep an endpoint entry relative to the immediate feedback from the rescue teams. These are images sent over when the teams are facing an issue, which are analyzed thanks to IBM Watson Image Classification technology.

Ultimately, AsTeR sees itself as an information collection platform going far beyond the analysis of emergency calls, for example displaying accredited sources of information such as government drones or satellites.

Thomas was previously a firefighter. How did that experience contribute to the project?

Thomas worked at the Paris Firefighter Brigade for 7 months as part of his military officer training. He led more than 300 rescue operations in one of Paris’ poorest neighborhoods. As a firefighter, Thomas witnessed first hand the distress of victims during emergencies. Interventions were varied and included assisting a mother in giving birth as well as rescuing a trapped driver in a car. Despite their diversity, all interventions shared a common difficulty: collecting information. Whether onsite or over the phone, gathering facts was often times the most challenging and time consuming aspect of the operation. This observation laid the foundations for AsTeR as an information collection platform.

How will your concept help cities respond to natural disasters or other emergencies?

During a natural disaster, fast and efficient collection of information saves time and lives. With AsTeR, victims will get help sooner and firefighters will be able to assist a larger amount of people in a limited time.

As an open source project AsTeR is highly modularizable. The platform can be adapted to include various sources of information. For instance, councils could create partnerships with local groups to get information from the ground when disaster strikes. News agencies could also become strategic partners for collecting information. Behind the scenes, having a real-time updated database will also greatly help the policy makers to take precise and evidence-based decisions in order to prepare for the next disasters to come.

As the go-to platform for emergency responders across the country, AsTeR would also greatly facilitate coordination of efforts during a natural disaster. When a natural disaster occurs, units from neighboring states may be called in to assist with the emergency relief effort. Fire departments from different states often use different radio channels and sources of information. This can slow down the deployment of forces and also lead to redundancy of efforts. AsTeR aims to make the whole process seamless.

During a natural disaster, AsTeR will therefore help cities make the most of their limited resources. Efficient collection, prioritization and sharing of information will facilitate decision making during large scale emergencies, ensuring those who need the most help get it first.


Learn more about IBM Code and Response.

 

Tagged

Related