• Interest in smart city technology • Flooding epidemic • Environmental Science Background • Natural Disaster • Shortest Path • Lack of access to vehicles within cities and for students

What it does

Smart Evacuate is a decision assistant that helps first responders and civilians coordinate quickly and avoid danger during times of crisis. This service dynamically tracks paths of disasters such as floods and fires; directing civilians away from areas affected by the disaster using the quickest route possible. If a route that was previously generated is later found to be inaccessible, the system reroutes the user in real-time to safety. In addition, the service provides a means for first responders to track the locations of users via GPS (which has accuracies up to 1 meter).

When a flood occurs, a Smart Evacuate sends out an emergency flood alert via SMS that gives civilians an option to provide a location to first responders if they are unable to safely evacuate due to a lack of transportation. It also provides a real-time map for civilians to route their own evacuation.

In the case that civilian are not able to evacuate, they would provide their location to first responders. At the time Smart Evacuate would populate the first responder’s map and route according to prioritizing civilians that are more likely in danger and to provide first responders with data on distributing resources.

How I built it

The basic premise of our application is to provide support services for those caught in natural disasters, specifically flooding. This product allows emergency response teams to effectively: 1) map out edges of a flood zone, 2) locate and route to individuals in need, and 3) provide dynamic routing around flooded areas. We start by using Fleetilla’s fleet services API to manage a fleet of support services. Their API allows us to track vehicle GPS location, speed, and direction, allowing us to route teams more effectively. We implemented a basic python wrapper for the Fleetilla API. We then integrated this API with the Google Maps API, which allows us to visualize and route between both individuals in need and support services. We host a static webpage (using cherrypy) containing an embedded, javascript-compatible Google Map. Using cherrypy, we are able to communicate across named micro-services to different devices in the network. We also built a machine learning detector to detect flooding in road images. With a Fleetilla-enabled car with an accessible dash cam, the car is able to detect areas of flooding and report them to the server to be shown on the Google Map.

Challenges I ran into

Most of us had very little (or no) experience with Javascript and other related web development tools. Therefore, we had to learn these on top of doing our project.

Accomplishments that I'm proud of

We sampled topographical maps to gather elevation data, then interpolated the elevation data into contour lines. While developing, we used these to predict the flow of a flood around Ann Arbor.

What I learned

We learned a bit about semantic segmentation and there use with image analyzation, as we used these to simulate a camera viewing a flooded area.

What's next for Smart Evacuate

Provide these services to Ann Arbor. Then move on to larger cities prone to disaster and then create the next SMART CITY.

Share this project: