Why we built it
In September 20, 2017, Puerto Rico was devastated by Hurricanes Irma and María. Many people lost their homes or died from not being able to access medical assistance. It was specifically catastrophic for communities in remote areas. First Responders are a great resource to assist communities right after a natural disaster, but in complicated geographical regions like Puerto Rico, it can be hard to reach isolated mountainous areas. We decided to develop disCOM to provide another resource for these communities who may not have access to communication or a first responder visiting them.
What it does
disCOM is a Raspberry Pi assistant that will communicate the needs of communities to local authorities. It will ask the community representative about what specific needs they may have after a natural disaster, such as: Food, Water, Medical Assistance. It uses on-device speech processing through the IBM Watson Assistant and Node-Red.
How we built it
The device is a Raspberry Pi 3 module running Raspbian. Node-Red was installed to provide the local hosting for the IBM Watson Assistant. The IBM software provides the text-to-speech interface and the speech recognition processing. Input text gets analyzed to obtain the necessary data from the user and then send the output to a server.
Challenges we ran into
Developing this idea with limited hardware resources was a complex task but possible.
- The Raspberry Pi had to be configured headless and without a WiFI connection since the WiFi on the hackathon venue was overloaded with network traffic.
- The low network speeds also made any downloads take an extensive time, which limited maximum productivity at times.
- Initially we had started with a Raspberry Pi Zero but its processing power was very limited, so we had to transfer everything to a Raspberry Pi 3 with an OS that includes a GUI.
Accomplishments that we're proud of
We were able to develop a device that could provide a natural language interface and process the community leader's responses. It is extremely efficient and processing runs entirely on-device (with the exception of data transfer). It is a very cost-efficient model that could be implemented at large scales to provide that needed communication beacons for people in remote locations to voice their needs. The potential of a device like this is astounding.
What we learned
- Everything about Raspberry Pis!
- The many innovative features included in IBM Watson Assistant
- Natural Language Processing Alternatives and how to make them run on different operating systems
- Text-to-Speech Engines
What's next for DisCom: Your Smart Disaster Communications Assistant
Building a RasperryPi Peer2Peer mesh network where info can transcend through the network through radio waves up until it reaches a device that has internet connection, which then sends the info the server to show to the authorities. We hope disCOM can really help people who are affected by natural disasters to get the resources they need, no matter where they are.
Log in or sign up for Devpost to join the conversation.