We all know to take weather forecasts with a grain of salt. Weathermen can tell us what's coming with pretty decent accuracy a day or two in advance using a bevy of sensors all over the country — as well as information from special, storm-hunting aircraft — but 10-day forecasts are little more than educated guesses.

So how do we improve predicting sunny or stormy days? Those beautiful minds over at MIT's Department of Aeronautics and Astronautics, led by principal investigator Jonathan How, think that robots are the answers. A swarm of unmanned aerial vehicles, such as the Boeing ScanEagle pictured above, could be equipped with sensors that would help weather forecasters get a better sense of the bigger picture when predicting the weather for the country. The models we use to forecast are pretty good, but, if we don't have all of the appropriate data to begin with, they tend to miss the mark.

The Boeing ScanEagle is showing promise as a storm hunter because of its relatively large size and weight, and its capability to fly 1,000 miles even with a full load.

Department of Aeronautics and Astronautics, via Technology Review