AI for the Common Good: F’AI’R Education Hackathon

This post has been republished via RSS; it originally appeared at: New blog articles in Microsoft Tech Community.

Winners AquaSpark - Water and Energy Management for All

 

The Team We are Dao Heng Liu, Vincent Leong, Rikaz Rameez and Joanne Wong, and we are second year students studying Computer Science at University College London.

 

 

The university is full of opportunities, and when we saw one such opportunity in the AI for the Common Good hackathon, we couldn’t pass it up. With a team being proficient in backend technologies and adept front end skills, we wanted to write something that would incorporate both ends and would work as a functional product. Driven by the promise of opportunities, prizes, and free food, we decided to attend the hackathon and try to write something useful, unique and realistic.

The Data

Data and examples were provided using Azure notebooks, access and links are up on x5gon.orgclick the hackathon link

The Problem

The hackathon has three overarching themes available to us.
They were:
- Supporting NGOs and NCOs
- Engineering Education for all.

- Health Education for All.

There are also three separate subtopics under each of the main themes, and one particular topic

-- Teaching Water and Energy Management -- took our interest. Being students who recently started to get our own utility bills, the problem of managing water and energy has become a personal issue for us. Previously we are very unaware of how much water and energy we use each month, and we have also discovered that we are not alone in having that issue. People around the world are generally unaware of how much energy and water they use, and how their usage impacts the environment, which is particularly important when facing the crisis of climate change. There are also associated costs of using such utilities, and people are unaware that reckless usage of electricity would entail exorbitant bills. This poses a problem especially in LEDCs (Less Economically Developed Countries), as they do not have as robust of a network compared to more developed countries. Careless usage might result in strain or even failure to the network, threatening the already fragile networks that LEDCs might have.

However, water and electricity meters could be hard to read, and even in the developed countries the installation of such meters are not universal. Many would like to monitor their usage statistics but have found no convenient way of doing so.

To address this issue, we would wish to not only allow users to monitor and track the water and energy usage of their household, but also to provide additional education materials in the form of videos and reading materials to teach the user about responsible use of utilities.

 

 

Our Idea Initially, we decided to create an water/energy management phone application where users can connect IoT devices to the app to monitor their water/energy usage. The application collects resource usage data over time and provides data visualisation as well as analysis. Users are able to set themselves a monthly budget of their preferred spending and receive notifications when they are reaching their budget limit.

 

After several discussions, we agreed to create an application that can be used on all platforms, be it on a desktop or a mobile device as part of our take on responsive design. The application has several features, including:

● Daily usage data visualisation
● Historical usage data visualisation + analysis
● Video recommendation for education purposes
● Article/reading material recommendation for education purposes
● Provides insight into sources of energy usage Utility usage is collected periodically and is graphed on a daily, monthly or yearly basis, depending on how the user chooses to view their data.

 

Using our API and depending on the users’ device setup, we would be able to calculate the utility usage for each individual appliance, which would give insight to the user about resource usage by appliance and would allow the user to identify points where they can improve their usage.

Technical Implementation

The web application is the front end of the implementation. In order to quickly develop a usable interface for the user and start testing the features, we opted to use an open-source bootstrap template as a base and build our implementation upon it. The web application includes a dashboard which displays historical data stored in an Azure SQL database, which is also accessible through an authenticated API. The web application itself is hosted using Azure’s Web App to allow for easy worldwide access, as well as to easily deploy webapp updates.

 

Further insights and statistics such as itemised utility usages and usage history would also be available as data stored within the databases. To inform the user about responsible usage of water and energy usage, we would also provide some audio and video information that is sourced from the open source X5GON API. The collected usage data would aid in the selection of relevant material to provide the greatest help for all different kinds of users.

 

We chose to use an Azure SQL database because it allows us to easily scale our application without having us to manually manage it while also ensuring high availability and uptime. All the while, the database meets the most stringent data protection policies with industry-leading information protection. One less thing to deal with!

 

 

Project Next steps

We aim to continue working on AquaSpark. We plan on extending our application by developing powerful new ways of plotting and viewing utility usage, and perhaps add more utilities such as gas.

 

We intend on developing a standard data format for our ingestion engine so that any user can feed data into the application. Developing a standardized data format is more feasible than using IoT device as the devices would need to undergo testing, certification and regulation. This results in increased cost both for us and the end user. With consent from the user, utility companies can send their data on their customers behalf, however users are free to install their own IoT/DIY devices capable of reading meters and connect the device to the platform to feed their own data.

 

We intend to implement machine learning algorithms that would analyse a user’s resource usage and suggest relevant videos, readings or tips to help the user to improve their resource usage. They should also alert users whenever atypical usage is detected, such as when there’s a leak, or they didn’t fully close the shower water. The algorithm would be initially trained on previous activity from other users to be used as an indicator for typical user behaviour, and then be periodically retained in order to take into consideration the user’s trends. The algorithm should also suggest tips on how to reduce consumption, relevant to the user’s usage trend. In the long run, we could also implement machine learning/data mining algorithms to study regional behaviours, as well as specialised displays designed to analyse regional or broad-area information.

 

In order to handle large amounts of traffic, we will pair our web app with Azure’s Front Door (load balancer and Web Application Firewall (WAF)) and Content Distribution Network (CDN). This architecture, in combination with Azure’s SQL database that we currently are using, ensures that our web application stays responsive, available and safe from common attacks.

 

One possible future addition is to add a social media aspect to our application, where our users would be able to compare usage, suggest videos and tips that would help others improve their usage.

 

We hope we will be able to make some significant progress in our project and that we will have a submission that stands a chance in the final round judging of the hackathon.

Next Event

AI for the Common Good: F’AI’R Education Hackathon 25-26 February 2020 Paris AI for the Common Good: F’AI’R Education Hackathon. This two-day hackathon hosted by the British Embassy in Paris, with the support of The Family, will see undergraduate students from four leading European universities – Université de Nantes (France), Universität Osnabrück (Germany), Ljubljana Jožef Stefan Institute (Slovenia) and University College London (United Kingdom) - compete in an Open Education & AI challenge. The aim is to bring together minds and ideas to build and deliver innovative and impactful AI-driven solutions to real world issues such as access to education and healthcare, capacity building and knowledge exchange in developing countries. During Day One (8:30-16:00), students will compete in an Open Education and AI challenge that will be presented back to the judging panel comprising AI and Ethics experts on Day 2 (8:45-15:00) including the Microsoft.
Panel of experts:

  • Ha Cole – CTO, Microsoft Philanthropy (Industry panellist)
  • Dr Husna Ahmad OBE – ANCSSC (NGO panellist)
  • Dr Louisa Zanoun – Senior Science and Innovation attache, British Embassy, Paris (Government panellist)
  • Sasha Rubel – AI Programme Specialist, UNESCO (International panellist)
  • Professor Kai-Uwe Kühnberger – Vice President for Research, Professor in Cognitive Science, Universität Osnabrück (Acadaemic panellist)
  • Davor Orlic – COO Knowledge 4 All Foundation (X5gon)

The judging panel will be looking for a polished product and pitch. The projects will be judged based on your team’s ability to meet user needs and develop a functioning, responsive API. Look forward for an update on the outcomes of the Paris hack but this process has been a great chance for students and academics to network with international colleagues.

REMEMBER: these articles are REPUBLISHED. Your best bet to get a reply is to follow the link at the top of the post to the ORIGINAL post! BUT you're more than welcome to start discussions here:

This site uses Akismet to reduce spam. Learn how your comment data is processed.