We operate drone delivery for medical supplies in rural Ghana and Kenya. Our fleet of 15 drones makes 80+ deliveries per day. Currently routes are planned manually by a dispatcher.
We want to automate multi-drone route planning — given a set of delivery orders, weather data and battery constraints, find the optimal assignment of orders to drones and the optimal flight path for each.
This is a combinatorial optimisation problem. We're open to RL, heuristic approaches, or hybrid methods — we care about solutions that work in the real world, not the theoretically optimal approach.
Requirements
– Combinatorial optimisation or multi-agent planning experience
– Python, simulation environments
– Familiarity with VRP (vehicle routing problem) variants is strongly preferred
– Comfortable iterating with an operations team who will tell you when the model is making bad decisions