Since the introduction of ROVs and other subsea vehicles in the 1980s, subsea robotics have undergone changes in software, tools and build, yet many feel there is still much to improve. Experts at CERAWeek by S&P Global said artificial intelligence (AI) looks to be the key to unlocking the full potential of subsea vehicles.

Subsea robots have come a long way, noted James Bellingham, executive director for the Johns Hopkins Institute for Assured Autonomy, at the “Robotics and Drones: Applications in Automation and Safety” panel on March 20.

“Everything was made from scratch, including the software,” he said. “No GPS, no satellite comms. If you wanted a navigation system, you built it yourself. Our batteries didn't last very long and they frequently caught fire… We weren't really sure who we added value to. We didn't really have a clear picture.”

Nowadays things are much different. While improved navigation and better batteries are nothing to sneeze at, the biggest addition from subsea vehicles has been the subtraction of people. Reducing the number of people offshore and keeping them out of harm’s way added value to ROVs, drones and other subsea vehicles and has been the main driver for innovation in the space.

“The driver really is trying to reduce the number of people offshore… we’re trying to reduce the size of boats, which reduces emissions and reduces the number of people on deck, which reduces safety hazards and clears people from the red zone,” John Gibson, CEO of Nauticus, told the audience. “Our whole vehicle emphasis is on reduction of emissions and reduction on people that are operating on decks so that you can make it a safer environment.”

With AI addressing safety concerns, robots and subsea vehicles are now optimized for their main objective—collecting data, said Troy Demmer, Chief Product Officer of Gecko Robotics. Using robots to collect data allows for more information to be gathered while lowering upfront costs for a higher return on investment.

AI also gives operators new insights, allowing them to identify and target more efficient mechanisms to manage assets over an entire lifecycle, Demmer said.  

“We're talking repair cycles, we're talking replacement and stretching out that predictive maintenance cycle,” he said. “This is enabled by software and AI that can power churn through millions and millions of data points using algorithmic detection and machine learning to extract those insights and serve that to the customer in a way that they can take action.”

Now, the next step is 100% autonomy—using AI to receive a mission, collect data, craft insights and then transfer it to the cloud without ever needing human intervention. But operators may want to pump the brakes for a second, Gibson said.  

“You'll know that humans are ready for autonomy when you get into a Tesla and there's no steering wheel there,” Gibson joked with the audience. “While there is a steering wheel there, there's some desire to have control to reach and grab it and give it a good turn once in a while.”

Gibson said to exercise caution with full autonomy; the amount of data from autonomous vehicles is not enough to accurately quality control a robot. Operators must also be “comfortable with the fact a vehicle works and acts on its own,” he said.

Subsea robotics have come a long way since first being rolled out for use in the oil and gas industry in the 1980s. And while there is still a ways to go, AI looks to close that gap.