Deep ocean robotics is not generally an area where we expect to see much in the way of significant innovation. When we do write about submersible robots, they’re usually confined to very near-surface operations. This isn’t a total surprise: It seems like the only people who really worry about what’s going on in the deep ocean (meaning hundreds or thousands of meters beneath the surface) are the military, the occasional scientist, and the oil and gas industry. Robots are important to these folks, even critical in some cases, but the technology has been more or less stagnant for decades, which is why we don’t write about it very frequently.
To be fair, there are some very good reasons why it’s hard to innovate when it comes to submersible robotics. The environment is hostile, support is remote, and communication is a challenge. Sound like any other environment where there has been a lot of robotics development lately? Sure it does: space.
Houston Mechatronics Inc., a robotics startup based in (you guessed it) Houston, Texas, has just announced an impressive $20 million Series B funding round. You may not be familiar with Houston Mechatronics itself, but you’re almost certainly familiar with some of their previous work: HMI was founded in 2014 by a team roboticists from NASA’s Johnson Space Center, including CTO Nic Radford, who was the chief engineer on the Robonaut project and led the group that developed Valkyrie.
HMI is targeting their decades of robotics experience at (among other things) a new generation of minimally supervised unmanned undersea vehicles (UUVs) that can do complex manipulation at depths of up to 3,000 meters without the need for support vessels. Called Aquanaut, HMI’s robot will travel in submarine mode to its deep water destination, where it’ll unfold itself into a sort of humanoid torso with arms and a head (!) and use a hybrid of remote control and assistive autonomy to get its work done before autonomously returning home.
If you want to do work at a substantial depth in the ocean, the established approach is to send down a remotely operated vehicle (ROV)—a boxy-looking miniature submarine thing with some thrusters and manipulators. ROVs have little or no autonomy, and need to be tethered to the surface for power and control, which means that you also need to send a sizeable support vessel packed full of highly qualified humans out to wherever you want to do any work. This gets very expensive very quickly, but it’s just the way things have been done. And since ROVs are working within an industry dominated by a few giant companies with enormous amounts of money, the incentive to find ways of doing things cheaper or more efficiently hasn’t been all that high.
A computer-rendered image of Aquanaut in “submarine mode.”Image: HMI
Houston Mechatronics wants to disrupt this space with a UUV that’s designed from the ground up (or is it from the waves down?) to operate in some fundamentally different ways than existing ROVs:
- Existing ROVs need large support vessels for deployment and operation. Aquanaut can be dropped off with a small boat (or even dropped from an autonomous helicopter), and then it swims to its final destination by itself.
- Existing ROVs need to be tethered for power. Aquanaut is fully electric and carries its own power supply, with >30 kWh of batteries onboard.
- Existing ROVs need to be tethered for human control. Aquanaut communicates acoustically through the water, which places severe limits on bandwidth (a few tens of kilobytes per second, at best). To compensate for this, Aquanaut will incorporate sophisticated autonomy that is completely new to UUVs.
Of these differentiators, we think the last one is both the most revolutionary and the most technically challenging. HMI is proposing to operate Aquanaut in an environment where degraded communications (like robots experienced during the DARPA Robotics Challenge Finals) are just the normal state of operations. And as with the DRC, some amount of autonomy, whether it’s full autonomy or assistive autonomy with a human in the loop, will be necessary.
“One of the particular skills that we had developed at NASA was centered around this major concept: dexterous mobile manipulation in really crappy communication environments,” HMI co-founder and CTO Nic Radford tells us. “In spaceflight, communication to remote assets like robots can be fairly intermittent, slow, latent, and lossy. Just like space, remotely operating robots underwater involves dealing with major uncertainties in the robot’s action. Therefore, optimizing the transfer of information from operator to the robot for action and having the robot understand the intent of an operator’s command is critical. The robot can then plan that action, execute it, and confirm its success back to the operator.”
To make this work, Aquanaut will be piloted with minimal supervision, using what Radford calls “a framework of autonomous function building blocks.” Our guess is that this means operators will be able to send high-level commands like “turn this valve 360 degrees clockwise,” without having to micromanage the movements of Aquanaut’s actuators. Getting this technique to work reliably has implications beyond just undersea robotics: There are all kinds of situations where communications are problematic, from controlling drones over long distances to robots in disaster areas to a future of robots doing complex manipulation tasks on the Moon and beyond.
There’s only so much that we’re able to tell you about Aquanaut itself at this point. Officially, a mockup of the robot will be unveiled at AUVSI in Colorado next week, but we’ve seen photos of bits and pieces, including working arms along with a partial body. Aquanaut will also be putting on an appearance at ICRA in Australia next month, where we’ll hopefully be able to get deeper into the technical details of the robot. For now, we have more on Houston Mechatronics itself, from Nic Radford.
IEEE Spectrum: How did Houston Mechatronics get started, and what was the transition from NASA to robotics startup like?
Working for NASA was an honor for me and others at our company, and some of the most formative years in any career one can hope to have. Developing robots for NASA’s missions afforded us an opportunity to equip ourselves with skills that we felt had tremendous commercial value and we chose to capitalize on that. HMI was formed when we decided we didn’t want to be sitting in a government lab watching the growing commercial robotics revolution from the sidelines.
Personally, I’ve found the transition to be invigorating. Starting a company is quite different than working for a government robotics research lab, but probably the biggest difference is the risk posture: startups are all about huge risks for huge returns. There is something so palatable about taking on the entrenched establishment in an industry, calling them out and saying, "I know you’ve been here for a while, but I think you’re doing it all wrong." There’s a reason NASA didn’t start SpaceX; often times established organizations don’t know of any other way to do what they do, and that’s what’s so thrilling about starting a new company— taking the risks.
What is Houston Mechatronics focusing on right now?
Through some robotic systems development projects coupled with a careful market study, we realized that operating subsea ROVs offshore is horribly expensive, and that technology in that area hasn’t evolved practically since it was introduced 30 to 40 years ago. So, we decided to formally take on the giants in this industry by redesigning the ROV with NASA-inspired command and control technology.
We’ve been developing and building a revolutionary fully electric underwater robot called Aquanaut for both commercial endeavors in the energy domains and defense related initiatives. On the commercial side, we’re building a robotics-as-a-service business around operating Aquanaut for tasks that are currently performed by ROVs attached to vessels or drilling platforms. On the defense side, the folks that perform the ’service’ are typically wearing camouflage, so we’ll gladly just sell them our tech. We’re getting support from the Navy and other folks from the DoD to help accomplish this. And did I mention that Aquanaut is also an underwater transformer? There’s that, too.
What other areas of robotics is Houston Mechatronics working on?
We are primarily focused on building Aquanaut and the full stack of related product suites (and proving some NASA-types can keep water out of a robot). For example, as part of our clean sheet approach to upending the ROV and AUV industries, we have developed all electric work class manipulators that could theoretically be applied to many subsea platforms. We also have our command and control software suite and electric thruster motor technology to highlight a few more things.
A computer-rendered image of Aquanaut in “humanoid mode.”Image: HMI
We mentioned above that Houston Mechatronics was founded in 2014— you may remember that the DRC Trials took place in December of 2013, and you also may remember that Valkyrie kinda didn’t do so well. Much later, we found out some of the reasons why, but it sounds like a big chunk of the NASA Valkyrie (and Robonaut) team decided to head out to do their own thing shortly after the DRC Trials, and that thing ended up as HMI— according to Randford, about half of HMI’s 45 employees came from the Valkyrie and Robonaut programs, which is a massive amount of experience and talent. We’ve been doing our best to keep track of these folks over the intervening years, and it’s super exciting that they’re (finally!) able to share what they’ve been working on this whole time.
For more on Aquanaut, check back with us during ICRA 2018, which is taking place in Brisbane, Australia from May 21st to 25th.
Evan Ackerman is a senior editor at IEEE Spectrum. Since 2007, he has written over 6,000 articles on robotics and technology. He has a degree in Martian geology and is excellent at playing bagpipes.