In order for drones to be used for delivery, they need to be able to deal with uncertainty in responding to factors such as high winds, sensor measurement errors, or drops in fuel. Such what ifs typically requires massive computation, which can be difficult to perform on a drone. Researchers from MIT have developed a two-pronged approach to reduce the computation. The first component is an algorithm that enables a drone to monitor aspects of its “health” in real time. This allows a drone to predict fuel level and conditions of the propellers, cameras, and other sensors. It also allows for proactive measures, if needed. The second component is a method for the drone to compute possible future locations before it takes off. This simplifies all potential routes a drone may take to reach a destination without colliding with obstacles. The algorithm uses the general framework of Partially Observable Markov Decision Processes (POMDP) to generate a tree of possibilities. The computation is separated into two parts: vehicle-level planning and health planning. As a next step, the researchers plan to test the route planning approach in experiments. The researchers have attached electromagnets to small drones to enable them to pick up and drop off small parcels.