The team has long experience from developing solutions that truly improve lives. Solutions from the indoor environmental perspective, that proactively improves workforces all the way from Helsinki to New York.

To truly challenge the team, there had to be something a bit bigger. But what is bigger than improving lives? Not much!

Except… Saving lives!

The situation in which we can truly save lives if of course emergencies. Often overlooked by many in tech simply due to how rarely they occur; evacuation procedures need to be improved by Big Data, and here was our chance.

What it does

There are two main parts of the solution. The first one is to make sure we are Always Ready, the proactive analytics. The second part is the application that helps when the crisis occurs.

The first part is the analytics, which combines multiples sensors such as Positioning, Cameras, Occupancy, Motion to analyse people-flow and identify changes in the environment and possible bottlenecks in the environment. In evacuation exercise situations it allows the user to see where things might have gone wrong, in a simulation. You see how long it takes for the building to be evacuated, and where mistakes are made, as well as if bottlenecks occur in the expected areas.

The second part is the application. This is an “every-man” application that allows anyone to become part of the rescue team, as well as avoid the bottleneck or crisis origin areas themselves. The application is an easily understandable interface that shows where there still are people in the building, as well as which floors have been evacuated. The evacuation still demands manual approval, as not to only rely on technology.

The application can be accessed on location by rescue workers as well, to better identify where to go next and what to do, based on where people are located in the building. The get a better understanding of where the crisis has originated, and are able to save lives more efficiently.

How we built it

We are using AWS (EC2 and RDS PostgreSQL database) to host our Frontend and Backend server. The backend has been implemented in Python (Flask-Restful). The Frontend has been implemented using PHP, HTML5, CSS and Javascript. The data analysis has been done in R & Python

Challenges we ran into

The lack of actual emergency drill exercises data was the major challenge. Ideation on the algorithms to extract each valuable piece of information were thought through but couldn’t be applied in practice on actual data. To overcome it, we ideated a lot on possible scenarios that could happen.

Accomplishments that we're proud of

We are most proud of the pace at which the multiple data source integration and analysis was able to be done. The solution can treat vast amounts of data and the algorithms can be modified when learning occurs, to enable a truly agile continuing development. The team’s diversity that could have been a challenge, turned into the real advantage.

What we learned

We learnt the basics of conducting an emergency drill and discover the need there is for occupants and building operators to better communicate on those matters in order to provide as much information and clear status to the fire brigade. Especially with changing working habits (working remotely…), it is difficult to appoint a permanent role to the occupants. Safety should be the primary concern of office workers, even if risks are remote.

What's next for “Evacuate!”

The prospect for our project is quite wide and full of opportunities:

  • Define best practices and KPIs for emergency evacuation drill exercises. We started by developing some metrics but the analytics of the results in the long run should provide a quantity of information necessary to create the best evacuation plan.
  • Increase the potential of usage of any available data source for incidents location (fire and others), record last locations of occupants to detect if they are in danger (in case of eventual system failure as a result of fire or other emergency situations)

As an elaboration on this, the co-operation with both authorities (fire brigade, rescue workers etc.) and building stakeholders would have huge implications from the Big Data perspective. The data collected in cases of emergency would be stored from each location up until a possible moment when the actual incident takes out the sensors.

  • This would give authorities and building stakeholders better information on how the smoke travels in the building and how and why some systems might fail in case of emergency

The case of emergency situations would also extend beyond the Finnish spectrum, and include states of emergency for e.g. natural disasters and terrorism.

Share this project: