Elderly and sick people isolated at home need an easy way to connect with volunteers for rapid assistance as online help-seeking is hard while frequent requests to relatives cause social embarrassment.
To address the problem and offer protection for the population in need during Covid-19 crisis, we offer COVER ― a service to connect people with COVid volunteERs via simple and centralized phone calls and sms.
Check our demo on YouTube!
Almost 20% of the Swiss population, belonging to 65+ age group, fall into a high-risk category according to Covid-19 regulations. While a lot of volunteering services and support groups are emerging on the Internet, only one third of senior people have a chance to reach out to their services since many still restrain from "going online".
Those who actually start searching for support on the world wide web may eventually get overwhelmed with an abundance of means and resources: numerous Facebook groups, WhatsApp chats, websites, mobile applications and many more... It is difficult to decide whom to reach out in this choice overload.
Some volunteers follow more traditional ways to suggest assistance for their elderly neighbors and leave paper notes on pinboards in shared hallways. However, one or two volunteers per building might not be enough to handle all the requests efficiently, not to say that fellow-neighbors may find it embarrassing to burden their younger or healthier counterparts too much.
The competitive analysis map (attached in the gallery) shows how we find our project different from other existing solutions. COVER platform aims to rapidly connect elderly or sick people to volunteers without a mediator in between ensuring ease of use and reduced social embarrassment.
This project aims to help elderly or sick people isolated at home and seeking for daily assistance (do grocery shopping, walk a dog, throw out trash, fix items, etc.). COVER is a platform to connect these people with volunteers in the neighborhood ready to assist. The service allows a user to request help via phone call, sms, or through the web interface, i.e. the means easily accessible to the senior population. The platform processes and sorts the requests according to their type. The volunteers available at the moment can accept a request and contact the person for further details. The requests are also displayed on a map for more convenience. The volunteer can receive notifications about new requests in the neighborhood on the fly, which facilitates and accelerates assistance.
To process incoming calls, we used the voximplant application, which connects the calling person with a dialogflow agent. The agent handles the call in natural language, retrieving the key parameters for the task: name of the person, phone number, type of service needed, and location. These parameters are passed to a Python-Flask backend server through a webhook. Once the server gets informed, it publishes the new task for volunteers on our interactive front-ends. We planned to have two types of front-end: web and mobile. We implemented a working prototype of the web interface using Vue.js and google map and geocoding api, while for the mobile interface we created an interactive mock-up with FluidUI. The idea is that any volunteer registered in the system can access any of the interfaces, accept the task, and call the person in need directly for further communication.
It is the first time we were dealing with processing phone calls. It was tricky to find a convenient communication API allowing to programmatically receive phone calls (we tried to use twilio first, but eventually ended up with voximplant). Configuring interaction between voximplant, dialogflow, and the back-end server was also a challenging task given our limited experience with this technology.
We are proud to be a part of the community fighting against Covid.
We put a lot of effort into making a minimal working prototype to showcase our project idea. We feel pumped up with what we managed to implement in just a couple of days: our first natural language phone bot, sophisticated interaction logic between applications and services, working web interface and nicely-looking mobile app mock-up (at least for us :) ).
Voximplant, dialogflow, fluidUI, Vue.js map
We participated in the hackathon mainly to share our idea with a broader community in order to attract professionals to take over the project and make it work in real life.
At the current stage, we have a minimum working prototype without an actual mobile app (only interface mock-up). So, the main further steps for the project would be:
- implementing mobile app
- configuring the phone bot to receive calls in different languages, depending on incoming phone number index (should be easily configurable with voximplant and dialogflow)
Link to the web interface: cover2020.pythonanywhere.com
Interactive app mockup: in FluidUI
To configure the call processing part for the prototype solution we used a test number provided in voximplant sandbox (expires 04.05.2020). To call COVER, you first need to call one of the gateway phone numbers and then enter our sandbox phone number as an extension.
Gateway phone numbers: +74999384362 (Russia), +19292240694 (United States), +48223970842 (Poland), +81345790139 (Japan), +97243720980 (Israel), +420228880669 (Czech Republic), +14388002812 (Canada), +61283104145 (Australia), +442038083060 (United Kingdom)
Extension: 699103171 (wait for the voice to end before entering that number)
Please, go easy on the entity recognition part :) (It works best with English, quality should improve when other languages are included)
Link to our DEVPOST submission
Short presentation in Google slides
Dialogflow:
Setup of Voximplant application:
- Building phone bot with Voximplant and api.ai
- How to use Dialogflow Connector (with Voximplant)
- Voximplant's Dialogflow Connector -> Setting Contexts
Communication of Dialogflow with your server using Fulfillment:
- Dialogflow Fulfillment: Webhook Tutorial Using Python (Flask)
- YouTube DialogFlow Tutorial - Fulfillment using a Local Webhook with ngrok
Other: