An article posted in the Polaris project webpage caught our attention in 2018, it was exposing the link between natural disasters and increased risk for vulnerable populations.

In Colombia, we suffered a major landslide in 2017 due to deforestation, trash accumulation in riverbeds and heavy rain. A doctor-without-borders report highlighted two (2) issues that we can feel can be tackle via technology:

  • No enough information about the extend of the natural disaster threat.
  • No real information about the population affected and vulnerable in the surroundings for first responders in the field.

What it does

Our ImaguDA solution uses a mesh of water-proof higrometer sensors on the field that detect soil moistures conditions that may result in a flood a determine the range of the possible impact.

The location of each sensor is shared with a serverless application that maps that information to census databases in order to list all the possible towns or cities affected and the total number of people vulnerable to human trafficking based in demographic parameters (age, gender and so on)

How I built it

Each sensor is attached to an IOT gateway built on top of a RaspberryPI 3 running the AWS IOT packages to connect to the IOT services via MQTT. Each gateway collects information about soil moisture and uploads it at configurable rates alongside update location calculations (coordinates) Once the AWS IOT service collects the message, an IOT rule is triggered to stored the data in a DynamoDB table for up to 7 days. A call to a set of lambda functions is triggered which:

  • Validates if the information from a each sensors can be considered as "pre-conditions for a flood" and estimates impact ranges as perimeters,
  • Queries census information web-services via API using the location reported by the an gateway and the perimeter defined.
  • Creates a summary of vulnerable population for each town or city in the given perimeter grouping data by age, gender, etc.
  • Triggers notifications to multiple field operatives via SNS. Finally, we feed the result to a ElasticSearch/Kibana cluster in order to present results and updates for Operations Center with a friendly dashboard which can enable better decisions to account for each potential victim.

Challenges I ran into

Mostly around data, getting accurate demographics for countries in Latin-America is tough. Our model was built based on commercially available APIs from ATTOM Data with accurate numbers for North-America.

One of our next actions is to secure similar metrics and indicators in services available for South-America including census information in Colombia, for example.

Accomplishments that I'm proud of

Being a contributor to the Doctors-without-borders NGO here in Colombia, its a great honor to feel that these initiatives can help Salaith Azuaque and other field operators to work with more information about the population affected. She was showcase in the 2017 Annual Report for here thoughts on the response to the Mocoa tragedy.

In these events, mental healthcare and overall aftermaths of the tragedies are usually overlook ... so it feels great to add a little contribution to prevent at least one incident of human trafficking under these conditions.

What I learned

The sheer volume of victims of human trafficking around the world and the scale of these events after natural disasters was a revelation. Being from Latinamerica, these events feel really close to us.

What's next for ImaguDisasterAlert

Creating an application that can enable first responders to add information obtained via field interviews and accounting for each of the groups reported in the notifications. In other words, putting a name and an identity to each of the vulnerable people reported.

Migrating to a more robust hardware platform to enable more cost-effective deployments in high risk areas.

Built With

Share this project: