The problem AQI.CO.IN solves

Problem Statement:

• Traditional climate data gathering methods are no longer scalable
• Lack of high-resolution climate data is hampering climate response
• Lack of reliable climate data poses a challenge for climate adaptation

Solution:

• Open-source climate data collection and incentivize participants
• Empower local communities to collect real-time high-res climate data
• Democratize access to climate data for an effective climate response
• Broader participation to achieve consensus and innovate on climate response
• Empower Climate Adaptation with Reliable and Standardized Data
• Optimize businesses and safeguard sensitive communities using micro-trends

Challenges we ran into

1. Building a low cost Open Source AQI Monitor requires a few trial and errors. We encountered challenges in measuring the PM2.5 density either due to a faulty sensor or lack of proper know-how using the sensor.
2. Getting the GPS module to work under the roof was a herculean task, but we seem to have achieved a location fix with the satellites.
3. Packing up these sensors and microcontrollers are still a challenge and we look forward to collaboration with others in the community
4. Storing climate datapoints on IPFS only returns a CID. However, this lacks enough metadata to stitch together a bigger picture. We had to develop a simple CRUD interface above the IPFS Shell to manage this in a RDMS manner.
5. On the deployment side, we faced minor tailwind configuration issues which was showing up in Vercel but not on Localhost. We solved it by using exportdefault instead of module.exports
6. To forecast the AQI data, we experimented with a few models like SARIMAX, Weatherbench2, and llama8b. We rain into some issues like inabilities getting real-time data. To fix this, we resorted to SARIMAX and checked for accuracy using SciKit-Learn.
7. We setup GaiaNet bot as a chatbot. But we were not able to receive reliable forecasts in realtime.

Additional Features

1. Fixing our old hardware setup and making them independent of (Serial) programmers
2. Placing some checks in the sensors before commiting datapoint
3. Moving the incentive module from RasPi to a central API
4. Developing a new Analytics/Leaderboard page ranking reward winners, device activity, and cities with cleaner air
5. Fixing our dashboard page (Removing Dark Mode and making the site further responsive)
6. Added a Chatbot interface in the dashboard powered by Gaianet
7. Development of our own forecaster model