Sean Halpin, Houston Mechatronics’ director of product management, explains that, for a long time, there’s been a “missing step” between AUVs and ROVs. “Standard AUVs can travel long distances to get to site, but they have no ability to reach out and touch, or manipulate the environment,” he says. “While ROVs can, they’re usually big fridge-like things, controlled by a surface-linked tether – so they can’t travel without a ship. We wanted to do both; the answer was something that could move from a swimming AUV to ROV form.”
As a travelling unit, the Aquanaut is about 3.5m in length and 1m high, with a familiar AUV morphology. But when the four linear actuators push the hulls apart, the unit opens up and becomes a 3m-high giant, exposing two additional control thrusters and the vehicle arms, and adding another degree of freedom to the vehicle head mechanism. “It really does look like a Transformer from the movie,” says Halpin.
Of course, producing the Aquanaut wasn’t easy, especially given that restricted timeframe. Halpin says:“But, the biggest challenge was trying to bring the vehicle to life, when it didn’t fit any existing stereotype. People know how to make AUVs and know ROVs, but not how to put both together. So we were breaking new ground all the time.”
Aquanaut’s alloy and composite body, bonded to buoyancy foam, holds a number of titanium vessels, charged to 1ATM, containing the electronics and computing ware. The twin arms have seven joints, ending in fully rotating grippers, yielding a high degree of dexterity.
There are several different types of sensors on board, including 3D-structured light, offset stereo cameras and sonar. While some will work fine in clear water, that’s not always guaranteed, so others step in to compensate, the ‘point cloud’ image being tied together by the software.
But sharing the data isn’t easy, as radio waves can’t penetrate water to any degree, so the transmission is acoustic and highly compressed – though “this is like dialling up an old modem, as it’s a lossy, low bandwidth, high-latency link”, says Halpin.
It’s not a problem, because the commands are more ‘directives’: therefore, Aquanaut doesn’t require real-time control. Halpin adds: “In practice it means the robot showing us the world as it sees it, and we’ll say, for example, ‘turn that valve’. It will coordinate all the arm motions by itself.” Just as well, as there are rather too many joints for an operator to handle.
But it takes one smart robot. As a result, there’s a lot of onboard software. “We’ve got three separate computers including one PC just for machine vision…in fact, the visual recognition is machine learning-based,” says Halpin, so it will go on accumulating experience as it develops.
In AUV form, the vehicle consumes about 500W to 1kW depending on sea motion and so on. While it’s a larger draw than a normal AUV, “it’s not so much... and we are in process of tuning it,” Halpin states. As a working ROV, this rate of energy consumption can rise to 2kW, plus there’s a significant power draw from the computing systems, second only to the motors.
There is 20kWh of onboard energy storage and, while this is a decent amount, it still demands a clever energy budget: “You can’t just throw all the power at the job in hand,” says Halpin.
Still, the team have balanced demand so Aquanaut can travel up to 200km in AUV form and still spend up to 12 hours working in ROV mode. But with luck, it won’t have to. Halpin explains: “We’re looking at our own flavour of vehicle residence - let’s call it ‘Uber for the ocean’. You park your robot in a predefined position, on a rig or maybe on a vessel of opportunity, and, when a job comes up, we can promise it’ll be there within so many hours.”