THE US military鈥檚 uncrewed aerial vehicles are a critical component of its search-and-destroy missions in warring regions. But could UAVs instead be used to save lives by flying search-and-rescue (SAR) missions in the nation鈥檚 large national parks?
, and colleagues at Brigham Young University in Provo, Utah, took a commercially available propeller-driven plane with a 1.2-metre wingspan, optimised for low-speed flight, and adapted it to both fly and search autonomously. Their idea is to provide SAR teams with a cheap alternative to helicopters, and one that can be used even in the perilous weather conditions that can ground helicopter-led rescue missions.
Without any piloting experience, an SAR operator can direct the plane to an area they want to search by clicking way-points on a computerised map, Goodrich says. The craft can also work autonomously to conduct search missions even if it is fed less specific information, such as a missing person鈥檚 last known position. To do so it uses computerised maps of the area to analyse the terrain and relies on probabilistic models to work out the missing person鈥檚 most likely routes.
Advertisement
The mistakes hikers make and the routes they take when lost are less random than you might think, says Goodrich. Topographical and environmental factors play a big role in determining where someone ends up. For example, once lost, people often stick to the easiest available ground rather than attempting to descend precipices or climb cliffs.
Experienced SAR team leaders already read the landscape to determine in which areas to focus the search effort and in what order of priority, says Ron Zeeman, a seasoned member of . Zeeman helped Goodrich鈥檚 team develop the algorithms that allow their autonomous UAV to do the same.
In trials, operators using the craft have taken between 35 and 150 minutes to find a dummy dumped in the wilderness 鈥� fast enough to impress Zeeman. 鈥淚f we could use it right now, I would,鈥� he says.
鈥淭he craft took between 35 minutes and 2陆 hours to find a dummy dumped in the wilderness鈥�
Although the system can predict where the dummy might be, the algorithms for detecting it within the images taken by the on-board camera still need work, says Goodrich. At the moment it鈥檚 usually the operator that spots the dummy first. With improved image-analysis software, the UAV could work entirely independently to locate missing people, he says.
The work will be presented at the in Atlanta, Georgia, this week.
