On February 6, 2023, an earthquake of magnitude Mw 7.8 struck southern and central Turkey, as well as northern and western Syria, with the epicenter 34 km west of Gaziantep. The earthquake caused widespread damage, leaving around 37,357 people dead and approximately 89,926 injured. At least 1.3 million people were displaced, and 24 million were affected. The earthquake also destroyed about 6,589 buildings and caused property damage of US$50-85 billion.

From the floods in Pakistan last year that left over 7.9 million people displaced and over 1700 dead, and the country still in shock, to the earthquake in Afghanistan to Hurricane Ian in the US that cost over $100 billion in damages, natural disasters have left the world reeling.

While they cannot be prevented entirely, it is possible and imperative to work on mitigating the destruction and loss of lives caused by such disasters. This includes predicting such events, being prepared for them, ensuring that the infrastructure is as resistant to them as possible, and being quick in response when such a calamity strikes.

The recent earthquake in Turkey is a tragic reminder of the importance of advanced technology in disaster management. It highlights the potential of AI to save lives and minimize damage.

How can AI help?

Drones for Construction Evaluation

One of the major causes of destruction during the Turkey earthquake was poor construction. Drones equipped with AI can be used to inspect building construction and notify officials about any substandard materials or faulty building structures. The BCA in Singapore is investigating the possibility of using drones to conduct building facade inspections. Using drones to evaluate building structures can prevent subpar construction that could lead to significant damage. Furthermore, drones can also be used for evaluating destruction and damage after an Earthquake has struck.

2. AI for Structural Assessment

AI can also assist in monitoring existing buildings for signs of damage that can compromise their structural integrity. This can ensure that all buildings conform to a minimum standard and undergo required maintenance.

Voltin, a Brisbane-based company and recipient of the Queensland Government’s Ignite Ideas Grant, uses artificial intelligence (AI) and machine learning (ML) to improve the accuracy and efficiency of building inspections. Their system, AutoBat2.0 (Autonomous Building Assessment Tool), uses geotagged RGB and thermal imagery to quickly identify potential defects and confirm structural stability. The identified defects are automatically mapped onto the building model and included in a digital building report. According to Voltin's Director, Stephen Thornton, their process is faster, cheaper, and more accurate than traditional methods, providing owners with a comprehensive building health check.

3. Systems for Early Warning

While earthquakes may still be difficult to predict, AI has proven effective in improving weather forecasting accuracy. For instance, Weathernews uses AI in Japan to collect and analyze real-time, hyperlocal data to provide accurate weather forecasts. In Thailand, AI warns factories outside Bangkok of sudden weather changes within three hours. By providing precise weather information, we can help communities better prepare for incoming disasters and potentially save lives.

Moreover, work is being done on AI-based Earthquake prediction software as well. A prominent example is DeepShake, a neural network developed by Stanford researchers to predict the intensity of ground shaking in an earthquake by analyzing seismic signals in real time. The model was trained on seismic recordings from around 30,000 earthquakes and used the earliest detected waves from an earthquake to make predictions and send alerts. It was developed using a university cluster of NVIDIA GPUs and was tested with seismic data from the Ridgecrest 7.1 magnitude earthquake, where it provided simulated alerts to nearby seismic stations 7 to 13 seconds before the high-intensity ground shaking arrived.

Another example is Mayday.ai - a startup based in Germany that offers an AI-based platform for real-time and near-real-time disaster and risk information services. The platform provides various services, including early warning alerts, geostationary and polar camera imagery, audio data, and social media content sentiment analysis. In 2022, Mayday.ai entered into multiple partnerships, including Airbus, Satellogic Inc., Picogrid, Overwatch Aero, and ConstellR, to enhance its risks and disaster intelligence capabilities.

When integrated with other disaster management measures, early warning systems can increase effectiveness, reduce panic, facilitate organized evacuation, and ultimately save lives.

4. Robots for Search and Rescue

AI-powered robots can support disaster response efforts by navigating tight spaces and communicating with each other to coordinate search and rescue efforts. Moreover, they can be utilized to look for survivors, transport necessary supplies, and aid rescuers. Let’s look at a few real-life examples of such robots.

Snakebot: Carnegie Mellon University's modular snake-like robot, which can adapt to different tasks, can inspect ships, submarines, and infrastructure underwater. The robot was deployed to search for trapped survivors in a collapsed apartment building after an earthquake in Mexico City in 2017. It provided rescue workers with a video feed from two passes through the rubble.

RoboBees, tiny Harvard-developed bots that use electrostatic adhesion to ‘perch’ on walls and even ceilings, evaluating structural damage in the aftermath of an earthquake.”

Such robots have the potential to save lives by locating and aiding people who may be trapped or injured in disaster zones.

4. Satellite-Based Damage Assessment

AI-based image and video analysis can facilitate the process of identifying damaged buildings and roads, which in turn enables emergency responders to focus on prioritizing rescue efforts. The United Nations Office for Disaster Risk Reduction (UNDRR) conducted a study that revealed that AI technology could accurately detect damage with over 80% precision using satellite imagery after a hurricane in Florida.

This technology can help save valuable time and resources by directing aid to the areas that require it the most, thereby enabling more efficient disaster response and recovery efforts.

xView2 is an open-source project sponsored and developed by the Pentagon’s Defense Innovation Unit and Carnegie Mellon University’s Software Engineering Institute in 2019. It relies on machine-learning algorithms that utilize semantic segmentation to evaluate satellite imagery and categorize building and infrastructure damage in disaster areas faster than current methods.

It was used during wildfires in California and for identifying damage after flooding in Nepal. It has also been used by search and rescue teams in Adiyaman, Turkey. While xView2 is up to 85-90% accurate, it has limitations, such as reliance on satellite imagery and the inability to spot damage on the sides of buildings.

Similarly, San Francisco-based startup CrowdAI uses satellite and aerial imagery to extract insights that aid disaster response efforts. The company has worked with telecoms provider WOW! to assess building damage after Hurricane Michael and with open data to analyze damage from the Camp Fire in Butte County. CrowdAI leverages NVIDIA GPUs in the cloud and onsite for AI model training and inference, providing real-time insights with just a one-second lag.

5. Using AI for Maintaining Effective Communication

AI can be a powerful tool to improve communication during earthquakes and other disasters. When disasters strike, communication networks are often disrupted, making it difficult for emergency responders and affected communities to communicate with each other. However, AI can automatically relay live updates and essential information to responders and residents in affected areas, ensuring everyone can access critical information in real-time.

The AI for Digital Response (AIDR) platform utilizes machine intelligence to automatically classify and filter social media messages related to emergencies, disasters, and humanitarian crises.

AI helps assess natural disaster risk and prioritize response by analyzing structured and unstructured data in real time. IBM's Operations Risk Insight (ORI) platform applies natural language processing and machine learning to visualize and communicate multi-hazard risks and was made available to non-profit organizations as part of the IBM Call for Code Program. IBM and NGOs have since partnered to improve and customize ORI for disaster response leaders. ORI provides hurricane and storm alerts, layered data sets, and map overlays to increase situational awareness for organizations such as Day One Relief, Good360, and Save the Children.

Social media monitoring tools powered by AI are essential in times of disaster. Dataminr is a company that offers personalized tools to extract relevant information from social media platforms to aid organizations, first responders, and governments in making informed decisions by providing current updates on aspects such as weather conditions, road closures, and power outages. Dataminr used its experience with AI models to work closely with UN Human Rights to understand its needs and develop a model to capture the information required to mobilize rapid responses.

In disaster response scenarios, AI-powered translation tools can be critical in overcoming language barriers between responders and affected individuals. While basic translation software like Google Translate can be useful, speech-to-speech translation applications like Meta's Speech Translator for Hokkien may prove even more effective soon. These advanced tools can enable more natural and accurate communication between responders and residents who speak different languages, improving the effectiveness of disaster response efforts.

Furthermore, AI can assist in “manning” emergency telephone lines. For example, in the past, IBM's Watson technology has been employed by the Association of Public-Safety Communications (APCO) to improve operations and public safety in emergency call centers. This initiative uses Watson's speech-to-text and analytics programs to listen to 911 calls. By utilizing Watson's speech-to-text function, the AI can analyze the context of each call and provide valuable insights that can help call centers respond to emergencies more effectively. This technology can reduce call times, provide accurate information, and accelerate time-sensitive emergency services.

The American military's Project Maven utilizes AI to analyze vast amounts of data and video footage captured by surveillance systems and alert human analysts of suspicious or abnormal patterns. While this technology was developed for military purposes, it has potential applications in disaster relief efforts.

By incorporating AI into disaster communication systems, we can improve the speed and effectiveness of our response efforts, potentially saving lives and minimizing the damage and destruction caused by earthquakes and other disasters.

Why Does this Matter?

Natural disasters and extreme weather events have become a major concern for the world's population. According to recent reports, over 1.81 billion people are at significant risk of flooding, and at least 3.3 billion people are highly vulnerable to the impacts of climate change. Furthermore, as the Intergovernmental Panel on Climate Change highlighted, the risk of death due to extreme weather is now 15 times higher than in previous years.

Integrating AI technology into disaster management is crucial to enhance emergency response efforts and assist affected communities. AI-powered tools can improve communication and coordination among agencies and responders, leading to more efficient and effective disaster response.

Furthermore, AI can also play a critical role in predicting the likelihood and severity of future disasters, thereby allowing communities and emergency responders to take proactive steps to mitigate their impact. Integrating AI technology into disaster management can help save more lives and minimize the damage and destruction caused by these catastrophic events. Therefore, we must continue to explore and invest in AI-based disaster management solutions.

AI has the potential to revolutionize disaster management, and governments and organizations must recognize this potential and take steps to incorporate it into their disaster management strategies. The world is vulnerable to natural disasters, and by leveraging advanced technology, we can reduce the loss of life and property damage caused by such events.

Revolutionizing Disaster Management: The Power of AI from Detection to Response