Last-mile delivery in East African cities operates under constraints that make the economics of conventional delivery challenging: traffic congestion that makes journey times unpredictable, road surfaces that accelerate vehicle wear, fuel costs that are high relative to delivery fees, and large portions of urban areas that lack formal addressing, making routing systems designed for Western cities unreliable.
This project is designing a hybrid autonomous delivery system that addresses these constraints directly. Ground robots handle dense urban corridors where traffic speed is low enough for safe mixed-use operation and pavement conditions allow reliable navigation. Fixed-wing drones handle peri-urban and unaddressed areas where road quality makes ground delivery impractical and aerial routing can bypass the addressing problem entirely. A software coordination layer routes packages between the two modalities based on real-time traffic conditions, distance, payload weight, and delivery time requirements.
The ground robot platform has been designed specifically for Kampala's road surface variability: the suspension geometry, wheel diameter, and ground clearance specifications are derived from empirical measurements of road conditions across the target delivery corridors rather than from generalised specifications. The mechanical design phase is complete.
The drone platform is in its second prototype iteration, optimising the trade-off between payload capacity, flight range, and unit cost at a price point that makes the delivery economics viable for the goods categories most commonly shipped in the target market.
The software coordination layer is functional in simulation across the Ntinda and Najjera corridors in Kampala. Current status: seeking a logistics partner for a real-world pilot deployment. The pilot scope would validate ground robot navigation performance on live road surfaces, drone delivery completion rates in the target corridors, and the coordination layer's routing decisions under real traffic conditions.
