Power to the People

About the project

Power to the People is finding rural homes in sub-Saharan Africa using satellite imagery to expand electrical access. To design electrical systems that meet the needs and desires of specific communities, we need detailed information about the places they call home.

Motivation

Close to 1 billion people live without electricity worldwide. Most live in sub-Saharan Africa or South Asia. The vast majority live in rural areas.

Electricity improves and saves lives. It allows for better healthcare through medical technologies. Electric night-time lighting allows for study and alleviates gendered safety issues. Electric cooking reduces harmful smoke and particulate exposure driven by fuel-burning cook-stoves. Productive uses of electricity, from electric grinding mills to sewing machines and refrigeration, can grow local businesses and alleviate poverty. There are so many ways electricity can improve livelihoods.

Our team is driven by the by the seventh United Nations Sustainable Development Goal, which calls for universal access to reliable and sustainable electricity by 2030. We want everyone to have access to electricity; in our modern, technology-driven world, electricity is essential in ensuring equal opportunity to thrive.

Achieving universal electrical access will require innovation. It will not be possible to provide universal access just by expanding existing national electricity grids. Reaching rural areas with national grid lines can be very expensive, both to the provider and the customer, and delivers more power than many rural communities initially need. It's just not a well-matched technology for the rural poor. Furthermore, as more and more distributed renewable generation resources (think of rooftop solar panels) enter national grids, even countries with well-established grids are experiencing grid control and management issues. It doesn't make sense to build outdated and expensive infrastructure to power the rural poor, who more than anyone need future-ready, affordable, and sustainable energy to climb out of poverty. We have to consider a much broader range of electricity options to meet the needs of these communities. This includes renewable off-grid solutions, such as solar home systems and community micro-grids. In selecting and designing these technologies, the wind, solar, and hydro power generation potential of the community must be carefully considered. Battery technology must be specifically selected and sized to the local context. By carefully matching the technology to the community, we can build energy systems that bridge the gaps unmet by existing grids.

Evaluating all of the factors which determine the optimal electricity technology for a given community can be a lengthy and expensive process. It can include multiple site visits to survey and map the community, and subsequent analysis of the collected data. This level of specific analysis is needed to ensure a community-appropriate grid is designed; however, these methods make it too slow to meet the scale of electrifying nearly a billion people in the next ten years.

We are working on tools to accelerate the design process for community-appropriate rural electrical grids. By extracting accurate and up-to-date rural home locations from satellite imagery, we can minimize the duration, quantity, and intrusiveness of preliminary site visits. We can arrive in communities with an idea of which electrical distribution topologies may best suit their community layout. We can even have preliminary grid designs drafted to enable easier discussions with community members and leaders. You can see an example of how this might look in a sample of Bidi Bidi in Uganda below!

That's why we're finding rural homes in satellite imagery: using these images of rural homes, we can train computer vision algorithms to accurately find homes on a massive scale, and use this information to design better electrical grids. This can allow use to accelerate grid design for rural areas while still tailoring grid designs to community needs. In turn, this can help to provide the essential service of electricity and improve the quality of life of the world's poorest people.

Goal

We aim to improve the accuracy of home identification computer vision algorithms for use in rural contexts and electrification. Available training datasets of home images typically do not represent the rural housing styles present in sub-Saharan Africa. We are developing a training dataset of rural homes in sub-Saharan Africa to train an object detection algorithm. This will be used to find homes in georeferenced satellite imagery, providing home locations to be used in electrical design. This method will enable cheaper and easier electrical system planning at the community distribution level, thereby allowing better electrical system planning for increased energy access. All research outputs including algorithms and training datasets will be made available for further research. We envision these being used by researchers and policy-makers to do system-level grid planning and prioritisation of electrification work.