A humanoid diving robot explores shipwrecks on the ocean floor

() — A robot created at Stanford University in California dives into shipwrecks and sunken planes in a way that humans can’t. Known as OceanOneK, the robot lets its operators feel like underwater explorers, too.

OceanOneK resembles a human diver from the front, with arms, hands and eyes that have 3D vision, capturing the underwater world in full color.

The back of the robot has computers and eight multi-directional thrusters that help it carefully maneuver around the fragile sunken ships.

When an operator on the ocean surface uses controls to steer OceanOneK, the robot’s haptic (touch-based) feedback system makes the person feel the resistance of the water, as well as the contours of the artifacts.

OceanOneK’s realistic tactile and visual capabilities are enough for people to feel like they’re diving into the depths, without the dangers or immense underwater pressure that a human diver would experience.

Stanford University robotics specialist Oussama Khatib and his students teamed up with deep-sea archaeologists and in September began sending the robot out on dives. The team just finished another underwater expedition in July.

So far, OceanOneK has explored a sunken Beechcraft Baron F-GDPV aircraft, the Italian steamship Le Francesco Crispi, a 2nd-century Roman ship off Corsica, a WWII P-38 Lightning aircraft, and a submarine named Le Protect.

The Crispi lies about 500 meters (1,640 feet) below the surface of the Mediterranean Sea.

“You’re moving very close to this amazing structure, and something amazing happens when you touch it: You really feel it,” said Khatib, the Weichai Professor in the Stanford School of Engineering and director of the Stanford Robotics Laboratory.

“I have never experienced anything like this in my life. I can say that it was I who touched the Crispi at 500 (meters). And I did it, I touched it, I felt it.”

OceanOneK could be just the beginning of a future where robots take on underwater exploration too dangerous for humans, helping us see the oceans in a whole new way.

Stanford University roboticist Oussama Khatib (second from left) was able to feel sensations in his hands using the robot’s feedback system.

Creating an underwater robot

The challenge in creating OceanOneK and its predecessor, OceanOne, was to build a robot that could withstand an underwater environment and the immense pressure at various depths, Khatib said.

OceanOne made its debut in 2016, exploring King Louis XIV’s shipwrecked flagship, La Lune, which sits 100 meters (328 feet) below the Mediterranean 32 kilometers (20 miles) off the south of France. The 1664 shipwreck remained untouched by humans.

The robot retrieved a vase the size of a grapefruit, and Khatib felt the sensations in his hands as OceanOne touched the vase before placing it in a retrieval basket.

The idea for OceanOne grew out of a desire to study coral reefs within the Red Sea at depths beyond the normal range for divers. The Stanford team wanted to create something as close to a human diver as possible, integrating artificial intelligence, advanced robotics and haptic feedback.

The robot is about 1.5 meters (5 feet) long, and its brain can register how carefully it should handle an object without breaking it, such as corals or sea-eroded artifacts. An operator can control the bot, but it is equipped with sensors and loaded with algorithms so it can work autonomously and avoid collisions.

While OceanOne was designed to reach maximum depths of 200 meters (656 feet), the researchers had a new goal: 1 kilometer (0.62 miles), hence the new name OceanOneK.

The team changed the robot’s body using a special foam that includes glass microspheres to increase buoyancy and combat pressures of 1,000 meters, more than 100 times what humans experience at sea level.

OceanOneK goes through an object grab test in a Stanford swimming pool.

The researchers enhanced the robot’s arms with a spring-and-oil mechanism that prevents compression as it descends into the depths of the ocean. OceanOneK also got two new types of hands and increased arm and head movement.

The project comes with challenges that it has never seen in any other system, said Wesley Guo, a doctoral student at the Stanford School of Engineering. “A lot of innovative thinking is required to make those solutions work.”

The team used the Stanford Recreational Pool to test the robot and perform experiments, such as carrying a video camera in a pen and collecting objects. Then came the ultimate test for OceanOneK.

deep dives

On a tour of the Mediterranean beginning in 2021, OceanOneK dived to these successive depths: 124 meters (406 feet) to the submarine, 334 meters (1,095 feet) to the remains of the Roman ship, and ultimately 852 meters ( 0.5 miles) to show that it has the ability to dive to almost 1 kilometer. But it was not a smooth road.

OceanOneK catches up with the cargo of an ancient Roman ship.

Guo and another Stanford doctoral student, Adrian Piedra, had to fix one of the robot’s disabled arms on the deck of their boat at night during a storm.

“For me, the robot has been eight years in the making,” Piedra said. “You have to understand how every part of this robot works: what are all the things that can go wrong, and things always go wrong. So it’s always like a puzzle. Being able to dive deep into the ocean and explore some wreckage shipwrecks that would never have been seen so close up is very gratifying.

Students work to fix a problem with OceanOneK during an expedition.

During OceanOneK’s deep dive in February, team members found that the rover was unable to ascend when they stopped to check the thruster. The floats on the communications and power line had collapsed, causing the line to pile up on top of the robot.

They were able to fix it and OceanOneK’s descent was successful. He left a commemorative marker on the seabed that reads: “A robot’s first touch on the deep seabed / A vast new world for humans to explore.”

Khatib, a computer science professor, called the experience an “incredible journey.” “This is the first time that a robot has been able to go that deep, interact with the environment and allow the human operator to feel that environment,” he said.

In July, the team revisited the Roman ship and the Crispi. While the first one has all but disappeared, her cargo remains scattered across the seafloor, Khatib said. At the Roman ship site, OceanOneK successfully collected ancient vases and oil lamps, which still bear the name of their maker.

The robot carefully placed a boom camera inside Crispi’s fractured hull to capture video of coral and rust formations as bacteria feast on the ship’s iron.

The robot extends a boom camera inside the hull of the Italian steamship Le Francesco Crispi.

“We go all the way to France for the expedition, and there, surrounded by a much larger team, coming from a wide range of backgrounds, you realize that the piece of this robot that you’ve been working on at Stanford is actually part of something much bigger,” Piedra said.

“You get a sense of how important this is, how novel and significant the immersion is going to be, and what this means for science in general.”

A promising future

The project born from an idea in 2014 has a long future of planned expeditions to lost underwater cities, coral reefs and deep shipwrecks. OceanOneK’s innovations also lay the groundwork for safer subsea engineering projects, such as ship, dock and pipeline repair.

An upcoming mission will explore a sunken steamboat on Lake Titicaca, on the border of Peru and Bolivia.

But Khatib and his team have even bigger dreams for the project: space.

Khatib said that the European Space Agency has expressed interest in the robot. A haptic device aboard the International Space Station would allow astronauts to interact with the robot.

“They can interact with the robot deep in the water,” Khatib said, “and this would be amazing because it would simulate the task of doing this on a different planet or on a different moon.”