Scientists will use 'bloodhound' and 'hummingbird' robots to plumb the depths of the Arctic Ocean, despite a huge risk. Anna Fazackerley reports.

The underwater robot had always come back before. But after it had dived beneath the Antarctic ice for a second time, there was total silence.

Having got into trouble, the robot tried to float to the surface. But with 200m of ice above it there was no escape. It would never be seen again.

Eight years after ocean scientists at Southampton University first dreamt it up, Autosub, their ?1.5 million unmanned underwater robot, was attempting the "Holy Grail" of missions. No other autonomous vehicle had ever explored beneath the Antarctic ice shelf.

ÖйúAƬ

Each mission was nerve-wracking. Once the sub had gone under, it was just a matter of waiting for it to return. "You can send commands to a spacecraft landing on one of Saturn's moons, but we have no such contact under sea water," Ken Collins, Autosub Under Ice's scientific co-ordinator, explains.

The project was so risky that it wasn't worth insuring their precious vehicle - it would be cheaper to start from scratch than pay the premium. On its first dive Autosub came up trumps, delivering images from a 25km journey under the floating ice shelf that would provide an important insight into the impact of climate change. But when the team sent the sub to retrace its steps, earlier this year, it all went wrong.

ÖйúAƬ

Collins is darkly humorous about the incident. "With climate change, at some point the ice will be gone and Autosub will be floating along. It still has our name on. If anyone finds it, we'd like it back please." Yet he admits that for the colleague who devised Autosub it was like "a personal bereavement".

Hanumant Singh, an American scientist who has collaborated with the National Oceanograhy Centre, Southampton, says: "It was terrible. But it could happen to any of us." Singh, an engineer at the Woods Hole Oceanographic Institution in the US, is building two vehicles for an international mission to the bottom of the Arctic Ocean basin. Autosub, which was looking up at the ice from underneath, remained relatively close to the surface. Singh's robots will plumb depths of up to 4,500m. The Arctic Ocean basin is tantalising because it is so inaccessible. It may hold more undiscovered scientific treasures than any other ocean basin - treasures that could provide new insights into the evolution of life. In 2001, scientists discovered possible hydrothermal vents along the Gakkel Ridge, the tectonic plate boundary that transects the eastern Arctic basin.

The speed with which plates move apart varies from place to place. It had been thought that in the Arctic they had been moving so slowly that there would have been little hydrothermal or volcanic activity.

"This has really sent biologists twittering," Singh says. "The race is on to find a hydrothermal vent, image it and analyse samples. Organisms in these vents may have been evolving in an isolated environment for 65 million years [when the basin became enclosed]. It is a bit like finding Australia." Rob Sohn, a geophysicist working on the project, sums it up:

ÖйúAƬ

"Fundamentally, we are on a voyage of discovery in what has to be one of the most remote regions on the planet."

Rather unpredictably, this has prompted $3 million (?1.6 million) of funding from space agency NasaLeague. Sohn says: "They see our using robots to find, image and hopefully even sample biological organisms of indeterminate kind and size in a remote and hostile environment as very much in line with their trajectory for planetary exploration."

Singh says that the project team has "a rough idea" of where the hydrothermal vents it wants to explore are. One "bloodhound" robot will be designed to sniff out changes in chemicals, temperature or turbidity that might indicate the presence of a vent. When one has been mapped, a "hummingbird" robot will hover in place taking images and collecting samples for analysis on dry land. The scientists agree that this is "by far" the most ambitious underwater project they have been involved in.

When Singh's creations set off under the ice in 2007, after millions of dollars of investment and thousands of hours of work, he will undoubtedly be recalling the recent Southampton disaster with a sense of panic. After five years of strategy meetings he is frank about the risk. "Under the ice there are huge implications to make sure the vehicle doesn't break. That probably means we would lose it."

ÖйúAƬ

"The Arctic ice pack may simply shut the door in our face," Sohn concurs, but he adds: "In a way, that is part of the allure."