High-flying drones and digital data crunchers are teaming up to find victims of the devastating earthquake in Nepal. These technologies give rescuers on the ground better information about where people may be trapped and how best to send supplies and first aid to remote villages beyond the reach of trucks or helicopters.
"What authorities are trying doing to do is to figure out what is going on," said Robin Murphy, director of the Center for Robot-Assisted Search and Rescue at Texas A&M University. "Then you have humanitarian relief organizations trying to meet individual needs on the ground. They are asking, 'Is this place flooded? What about roads and transportation here?'"
Murphy says unmanned aerial vehicles have been used since 2004 for eight previous earthquakes, mostly for reconnaissance and structural mapping of crumbled buildings.
In 2010, during the aftermath of the Haiti earthquake, the U.S. Navy also used underwater robots to clear the harbor in Port-Au-Prince so ships could bring in responders and supplies without running aground or destroying existing piers. The Chinese military used small UAVs to find victims of the 2013 Lushan and the 2014 Yunnan earthquakes, according to Murphy.
On Monday, Ontario-based Aeryon Labs dispatched three drone aircraft to Kathmandu to work with first-responders at GlobalMedic, a non-profit that provides drinking water and shelter to disaster victims.
"You can send them into areas that are inaccessible," said GlobalMedic executive director Rajul Singh. "If I can't get past the road I can put the UAV up there to see if anyone is there that needs my help. There are not enough helicopters in Nepal right now, and they shouldn't be taking pictures, they should be flying aid."
The Aeryon drones being sent to Nepal are equipped with thermal cameras to help locate survivors by detecting body heat, as well as the company's newest digital zoom camera that can see the details of human faces at 1,000 feet away. The team will also undertake aerial mapping of the affected areas, building 2-D and 3-D maps, for disaster teams.
Murphy said small flying UAVs have limitations. Backpack-sized drones are easy to carry into a disaster zone and to launch by hand, but they run on batteries and only have 30 minutes of flying time. With electricity to most of Nepal knocked out, that means relief teams have to haul in lots of spare juice.
"The trade off is that small means light," Murphy said. "They get buffeted by winds and eat up battery power. Operators on the ground need to think about weather and wind, as well as the impact of high altitudes on lift."
Another high-tech relief team is using a combination of digital crowd-sourcing and an artificial intelligence algorithm to scan images and tweets about the disaster to help relief officials on the ground. Patrick Meier is coordinating the effort from Doha, Qatar, where he is director of social innovation at the Qatar Computing Research Institute (QCRI).
He and others developed MicroMappers.org, a site where people can help sift through thousands of tweets and online images from Nepal to see if there is useful data for rescue crews. Each tweet and image is scanned and categorized by three different human users. The results are passed onto the United Nations relief agencies working on the ground.
So far, they've signed up 1,400 people in 70 nations to help with the effort.
"We are dealing with a giant stack of information," Meier said. "We have to sift through these haystacks to find actionable information that potentially is lifesaving. What we do is harvest and pull the pictures that are displayed in news articles and push that to the image clicker and asked them to be tagged by varying levels of damage."
The use of crowd-sourcing in identifying large numbers of digital images has been used before in other applications and other organizations to search for quasars in outer space, identify endangered species or even search through old ship's logs to find 19th century climate data. This time, the digital mapping project could make a difference for earthquake victims in Nepal.
Meier said his group recently developed an artificial intelligence platform that can automate and speed up the categorizing task of images, processing up to 2 million tweets per hour.
While that volume hasn't occurred yet from Nepal, it could make a difference for future rescue efforts in more densely populated areas, such as Los Angeles, New York or Tokyo - places not immune to large-scale disasters.