Âé¶ą´«Ă˝

Internet volunteers transform search and rescue

The internet-based searches for aviator Steve Fossett point to the future of search and rescue – and it's from your home PC

From my viewpoint above the Nevada wilderness I’m scouring the landscape for signs of a missing aircraft. I’m searching for Steve Fossett, the millionaire, record-breaking aviator whose plane went missing on the Labor Day holiday.

If it were any other search I’d be in the cockpit of a search plane, but this time it’s different. I’m sitting in front of my PC, 8000 kilometres away in London, sifting through satellite photos that I, like thousands of others, downloaded to take part in for the missing aviator.

As Âé¶ą´«Ă˝ went to press, Fossett’s single-engine plane was still missing, and the Nevada Civil Air Patrol had scaled down the search. But the photos scoured by online volunteers helped to reveal eight other, older aircraft wrecks that surfaced and would otherwise have been lost in the 44,000-square-kilometre area, highlighting the potential of online “crowd-sourced” search. My experience was almost certainly a glimpse of things to come.

The idea of calling for many people to join a common task is not new, but the internet has given it legs. “The internet is probably the only way you can do a massive search cost-effectively,” says Tim Ball, president of FireBall Information Technologies of Reno, Nevada, whose mapping work led efforts to find Fossett in the region of Minden, Nevada.

“The internet is probably the only way you can do a massive search cost-effectively”

The search for Fossett is only the second time online crowd-sourcing had been harnessed as a search tool, and both searches were only possible thanks to the powerful, high-tech contacts of the missing people. The first mission was mounted when Jim Gray, a senior Microsoft research engineer, disappeared in January after setting off from San Francisco in a 12-metre yacht. A sea and air-based search by the US Coastguard found nothing, but Gray’s tech-savvy friends were not prepared to give up.

“I’d met him a couple of times and a lot of us at Amazon and Google and Microsoft who knew him very well wondered what we could do as computer scientists to help find him,” says Peter Cohen of Amazon in Seattle, Washington.

They started by persuading their contacts at satellite operators Digital Globe and GeoEye, which Microsoft and Google use to create their geo-browsers, to donate satellite images of the area with 1-metre resolution, enough to spot a plane, car or boat, although not a human being. Then they sliced the images into segments and sent them to online volunteers using the website owned by Amazon.

Mechanical Turk

Named after an 18th-century chess-playing “automaton”, which was actually operated by a human hidden inside, Mechanical Turk automates the distribution of tasks that are difficult for computers but simple for humans, such as answering the question: “Is there a pizza parlour in this photograph?” E-businesses use it to search for suitable images for their websites, for example.

Photos of the stretch of sea where Gray had apparently gone missing were examined by the searchers. By making the images compatible with Google Earth, volunteers could tilt them in three dimensions. This made it easier to distinguish a plane wreck from something less interesting. The ones flagged as unusual by a majority of volunteers were prioritised for viewing by the physical searchers.

Cohen’s team learned which image resolutions worked well, how to refine the instructions given to inexperienced searchers, what size to make the image segments and how much the images should overlap each other. It also gave them a better understanding of which tips to follow up. Sadly, it didn’t help them find Gray, who is now presumed dead.

On 3 September, Fossett went missing after flying out of a ranch in Nevada on a solo jaunt. His friend and fellow aviation enthusiast Richard Branson appealed for help, and the high-tech alliance was back in action.

In the search for Fossett each picture covered an area 85 metres square and was sent to 10 people online. If the majority flagged it, the area was prioritised for viewing by the US Civil Air Patrol or Air National Guard teams, who conducted the physical search partly based on what looked promising from the photographs.

There are still problems with the technique. Police officer Russell Peterson of the Washoe county sheriff’s department in Nevada says that because the satellite images were taken after the air search had begun, some volunteers were flagging photos containing low-flying search planes that they mistook for Fossett’s craft.

The question now being asked is whether such an expensive search-and-rescue effort would be mounted for anyone less well connected than Fossett or Gray. “Why search for Steve Fossett and not just anybody?” Cohen asks.

Right now the main reason is the difficulty of finding a company willing to donate the pictures. “We are glad to donate some imagery sometimes, but finding missing people is not core to our business,” says Mark Brender of GeoEye in Denver, Colorado. “We would hope that there will in future be some way where either government or an NGO can purchase satellite imagery on a regular basis to conduct searches for missing people.” He notes that when nations want low-resolution images of areas stricken by floods, tsunamis, hurricanes or earthquakes, often pays.

Another problem with the satellites used by GeoEye and Digital Globe is that they are not guaranteed to pass over a given area of the Earth exactly when they’re needed. “Our Ikonos satellite can be overhead in any one area like the Fossett search zone once every three days,” says Brender.

If it’s a case of life or death, that might not be good enough. Although Cohen points out that satellite images are being taken more and more frequently, Ball has another option, and it’s already standing on the apron at Minden airfield, slung beneath an executive jet.

FireBall has teamed up with High Altitude Mapping Mission of San Francisco, which creates high-quality digital maps, to build a 160-megapixel aerial camera and image-processing system. Capable of imaging “state and nation-sized areas” from 20,000 feet, the aim is to overcome the limitations of satellites. The camera can provide 15-centimetre resolution as compared with the 41 centimetres provided by the newest commercial satellite WorldView-1, which Digital Globe . Also the jet can take off at short notice and be sent to cover a specific area. Even better for search and rescue, the images should be cheaper than those from satellites.

The equipment is currently being test flown. “We are going out to see if we can detect reliably some of the existing plane wrecks in Nevada,” Ball says. If it works, this could be the start of crowd-sourced rescue for all.

Aviation – Learn more in our comprehensive .

Topics: Aviation