Even with bottled oxygen and elite training, there are underwater locations that lie well beyond our physical capabilities. But via haptic feedback technology and artificial intelligence, Stanford University's humanoid diving robot is now putting the ocean's depths within human reach. In its maiden expedition, the OceanOne droid has just scoured an untouched shipwreck off the coast of France and returned with a delicate, 17th century vase in its grip. Researchers are now eyeing future voyages to coral reefs, oil rigs and underwater disaster zones.

With our deep sea diving capabilities only taking us so far, we have long sought to send robots down below to do the investigating for us. We have seen swimming robots inspired by jellyfish, eels, octopi and rays, but in the view of the Stanford researchers the human form could have plenty to offer these aquatic endeavors as well.

OceanOne was originally dreamt up as a tool to study coral reefs deep in the Red Sea with a delicacy of touch resembling that of a human. Around 5 ft tall (1.5 m), the robot is fitted with stereoscopic vision and a pair of fully articulated arms, while eight multi-directional thrusters, batteries and computers are built into its tail. Described as a robo-mermaid, OceanOne is controlled remotely using joysticks by humans sitting safely and dryly on a boat above.

"OceanOne will be your avatar," says Oussama Khatib, a computer science professor at Stanford. "The intent here is to have a human diving virtually, to put the human out of harm's way. Having a machine that has human characteristics that can project the human diver's embodiment at depth is going to be amazing."

Khatib was at the controls as OceanOne took its first dive. Its first port of call would be the shipwreck La Lune, King Louis XIV's vessel which came to rest 100 m (328 ft) below the southern coast of France in 1664. No humans had been able to reach the wreck previously, but with the help of earlier studies carried out by archaeological researchers at France's Ministry of Culture, OceanOne was able to successfully navigate the site.

Through the robot's eyes, Khatib saw a grapefruit-sized vase and reached out with the OceanOne's articulated arms to feel its contours and weight. This is made possible through force sensors that send haptic feedback back to the controls. The team says the feedback is so sensitive that the pilot can tell if the robot has grabbed something firm and heavy, or light and delicate. This combines with artificial intelligence to ensure that the robot holds the object firmly, but doesn't cause damage. The vase made it safely back up to the boat, where Khatib laid the first human hand on it in more than 300 years.

"You can feel exactly what the robot is doing," Khatib says. "It's almost like you are there; with the sense of touch you create a new dimension of perception."

Almost equally impressive is how the robot positions itself in the water. Using its sensors, OceanOne gauges currents and turbulence and automatically fires the thrusters to counter these forces and hold its place. And even if the body moves, the arms are automatically adjusted to keep the hands steady.

It finds its way around by learning about the environment through sensors and cameras, which collect data to be run through algorithms and avoid it crashing into things. If it senses an upcoming obstacle but calculates that it cannot slow down in time, OceanOne will use its arms to brace for impact, just as a human would.

But its navigational abilities aren't foolproof yet. As the robot approached the shipwreck it became lodged between two cannons. Unable to release it using the thrusters, Khatib took control of the arms and had it perform a push up, which in the end set it free.

In addition to providing a more intuitive underwater robot with a gentle touch, the Stanford researchers say that the human form also allows OceanOne to communicate with human divers using hand gestures. Despite this, they imagine it eventually diving with a team of other humanoids in fleets of robotic underwater explorers.

OceanOne will be returned to Stanford next month, where the team will continue working to improve on the technology. The overarching aim is to build a robot that can take on tasks that are either too dangerous or physically impossible for humans to handle. These might include studying deep coral reefs, underwater disaster areas like the Fukushima Daiichi power plant in Japan or maintaining oil rigs.

You can see OceanOne do its thing in the video below.

View gallery - 6 images