Powering sub-Saharan Africa beyond the grid

Solving last-mile logistics

Most rural communities are only connected by unmapped dirt paths across rugged terrain. To expand access to off-grid solar, the first step is pinpointing households that need it, and these homes don't appear in standard mapping apps. To locate these, our joint team combined satellite imagery and Google's Open Buildings dataset with smart routing algorithms.

The field teams then need to reach these remote households. Google researchers are developing advanced geospatial models that identify rural paths and incorporate them into real-world routing constraints.

Google's Operations Research (OR) models are helping to optimize Ignite's end-to-end supply chain. By evaluating potential network configurations, the system identifies where central warehouses and local mini-hubs should go while carefully balancing operational constraints such as transportation costs, delivery routes, vehicle capacities, and inventory holding expenses.

Scaling electrification from space

As electrification ramps up across Africa, operators need to know where micro-solar systems (usually consisting of a single rooftop panel) are already active. Mapping these installations allows teams to identify underserved villages, deploy resources strategically, and plan future rollouts.

However, standard satellite imagery only registers a single rooftop panel as barely one pixel, so visual targeting is nearly impossible at scale.

To solve this, Google researchers are developing computer vision models using Gemini for Science and fine-tuning them with ground-truth drone imagery provided by Ignite to map solar setups from orbit. The team is testing thousands of code variations to increase precision and recall rates, turning satellite data into a strategic compass for future clean-energy deployment.