In December 2013 Amazon announced that it would eventually start delivering packages by drone. In order for a drone to be capable of carrying out theses missions, however, it must be able to fly in an array of flying conditions and be reliable enough to courier expensive goods to and fro.
A new algorithm developed by researchers at MIT will supposedly significantly reduce the computation needed to complete long-range delivery and allow drones to monitor their own health. Drones will be able to keep an eye on their ability to do the job by predicting fuel levels and checking on the condition of propellers, cameras and other sensors. They will also be able to make decisions about how to best act on the information — rerouting to a charging station, for example. A second skill that drones of the future will be equipped with is the ability to efficiently compute their possible future locations offline, prior to take off. This will prevent it from colliding with other obstacles by potentially taking a different route to the one that has been planned. By planning these routes ahead of time, a significant proportion of the drone’s computational energy is freed up so that it is available to react while flying.
“With something like package delivery, which needs to be done persistently over hours, you need to take into account the health of the system,” says Ali-akbar Agha-mohammadi, one of the authors of the paper detailing the algorithm’s development. “Interestingly, in our simulations, we found that, even in harsh environments, out of 100 drones, we only had a few failures.”
Both algorithms were tested by multiple drones on simulated delivery missions and under an array of environmental conditions. Results showed that drones were able to deliver the same number of packages as those not programmed with the algorithms, but with a lower rate of failure and breakdowns.
From Around the Web