Inspired by the MIT course ”Artificial Intelligence: Implications for business strategy” I was thinking about the usage of Artificial Intelligence for Emergency Management (AI4EM) in the last 6 weeks. In this blog I would like to put some ideas up for discussion:
Emergency phone call inquiry
Inquiry of the initial call of an incident
The caller is often very nervous and therefore uses his native language. A rising problem in our globalized world.
A Nature Language Processing Tool is able to recognize the different languages and understands the important words (like fire, injuries, heart attack, building collapse), classifying the information (type, urgent, severe), transfer this classification to an dispatcher and finally answers the caller in his or her language.
If the NLP Tool is not able to handle the call adequately an human dispatcher take over the call. A Real-time conversational guidance application will support the dispatcher. This application analysis in real-time the mood of the caller (Sentiment Analysis) and provide the dispatcher with real-time feedback and next best action guidance. This increases the quality of the inquiry and decrease the stress for the dispatcher.
This module asks the caller about the nature and the location of the emergency and if this emergency is already reported it gives the appropriate answer and thank for the call. Otherwise it transfers the caller to the ”Multi-language Emergency Inquiry Module” or an human.
With this module the number of calls which has to be processes by the dispatchers decreases drastically. The dispatchers get more time for the important phone calls.
Answering ”Cat-up-the-Tree” Calls
Often people calling the emergency call center when they are in trouble and don’t know who is able to help them: e.g. cat up a tree, water pipe breakage, stomach upset, etc. In all these cases the AI-system answer shortly and forward the caller to the responsible organization.
Answering ”Request for behavioral informations during a crisis situation”
The information which the callers ask is usually limited. After a while the AI system has learnt the right response from the first answers by itself or experts have prepared standard answers which then the AI-system can use. Even instructions for cardiopulmonary resuscitation can be given by the system.
Monitoring Social Media for early detection
The AI-system is monitoring the social media, learning regular contents and patterns for the area of responsibility. If it recognize irregular patterns, e.g. if unusually amount of users are talking about restroom activities then the AI-system alerts an human analysts about a possible epidemic event. (see e.g. www.va-sa.net)
Dispatching units to emergencies
Dispatchers use the description of the event from the emergency caller for the composition the forces which should be alerted. Often they deploy too many or few forces. An Machine Learning tool learn from all former incidents and dispatches forces taking in consideration time, location, number of callers, state of excitation, weather conditions, season, etc..
Villa and Apartment Fire Fighting
Robots can be used in fire fighting mainly to increase the quality of the service. With autonomous turntable ladders and search robots the extreme critical time for search and rescue can be reduced. With each minute the surviving probability is increasing over proportional.
Autonomous Turntable Ladders:
Turntable ladders can be move technically in three dimensions simultaneously. This is a challenging job under stress. An autonomous system can take over this task (compare parking assistance). For recognizing the target one can mark it with a laser or use scene recognition technology.
A firefighter is driving the turntable ladder near to a building where people has to be rescued from windows. After recognizing the target the turntable ladder drive autonomously into the best parking position and move the ladder park.
Autonomous Search Robot for operation in a room filled with smokeAn autonomously moving robot is searching for humans being in a room on fire. It focus on places where especially children hide themselves. It uses sensors to detect if the person is still alive and mark the position with an electronic buoy for the firefighters who rescue the person.
The robots for room fires have to be able to get over obstacles which are typical lying on the ground in an household: toys, shows, etc.
Autonomous Extinguishing Robot
The task for the ”extinguishing robot” is to confine the fire. It drives directly to the heart of the fire and start firefighting there. By applying the extinguishing agents at the most effective location the whole process is faster and needs less fire extinguishing agents. This reduces the costs directly for the fire brigade and indirectly for the victims (owners, insurance companies) by reducing the secondary damages through the agents. Finally it reduces the lost for the national economy.
To ensure the collaboration between human firefighters and robots all important information from the robots are be shown in the visier of the firefighters’ smart helmets.
A secondary effect of using robots – but a not to be ignored one – is that the threats for the firefighters is reduced.
After-Operation Restore Readiness
Consumables of a fire truck has be refilled after an operation to get readiness again. These items can be ascertained directly at the scenery (using QR code system), transmitted to the station and a storage and delivery robot gather and bring them to the parking lot of the truck.
This reduces the time out of order of the trucks and therefore increasing the readiness of the fire brigade. The firefighters can use the saved time to clean themselves after an operation or for doing more drills and trainings. Both would increase the health care which indirectly reduce the costs for the fire brigade.
Sea Search and Rescue (S-SAR)
If a person missing in the sea a swarm of search drones scans a huger region. After getting first clues they change their searching strategy autonomously. If they located the person they send the position to an transport drone carrying a life vest and to an autonomous rescue boat.
Fire Brigade Business Planing
Forces Distribution Planning
On the strategic level machine learning will enables the authority to locate best places for stations and distribute the forces tailored to the requirements of the different areas of its responsibility. Goal is to minimize the arrival time for needed forces. Characteristics of events in a specific area and therefore the best distribution of forces depends on land-use, population, road infrastructure etc. and is changing over time. Machine learning tools recognizes such changes in a very early stage. With this early detecting of changes in the socio-cultural composition of the residents of an area the fire brigade is able to react to these changes before they becoming significant. Thus machine learning prevent quality decrease induced by changing population and urban development.
I am sure, that using AI can eliminate stress-related mistakes or at least noticeably reduce them. This results in more rescued lives and less physical damages. Additionally it reduces the live-threatening and the health risks for the firefighters.
Ministry of Interior UAE, United Arab Emirates, Strategic Advisor Department of Public Safety and Quick Intervention
Civil Expert for NATO Civil Protection Group
Geboren 1962 in Braunschweig. Ich arbeite, unterrichte und forsche seit einigen Jahren in den Bereichen Disaster Response, Entscheidungsfindung, Stabslehre und Spezielle Einsätze im Bereich des Bevölkerungsschutzes.
Seit meinem 16. Lebensjahr bin ich Mitglied der FF Braunschweig/Rautheim und seit 2004 im THW Bochum.
Nach meinem Studium der theoretischen Kernphysik an der Technischen Universität Carolina Wilhelmina zu Braunschweig absolvierte ich mein Referendariat zum höheren feuerwehrtechnischen Dienst bei der Berliner Feuerwehr. Danach arbeitete ich als Direktionsdienstbeamter bei den Berufsfeuerwehren Stuttgart und Bochum bevor ich 2006 als Lehrbereichsleiter zum BBK an die AKNZ wechselte.
Seit Mai 2014 arbeite ich für das Innenministerium der Vereinigten Arabischen Emirate.