This article is more than 1 year old
Puny human sailors still needed... until drone machine learning tech catches up
RN admiral insists robots won't replace sailors, for now
Drones won’t replace proper sailors anytime soon because, believe it or not, they need more manpower to operate, a Royal Navy admiral has insisted.
Naval drones are “not about reducing the requirement for people”, Rear Admiral Paul Bennett told a press briefing attended by El Reg on Friday. Instead, they are for putting people into positions where they add “real value”.
At present, unmanned systems - drones - require on average something like four or five operators each, we understand. Rather than enabling cuts in manpower, if anything they require ever more personnel aboard ships to operate them; not a good situation to be in when the Navy is already critically short of heads.
“If you look at what's happening in the mine countermeasures world, for example,” continued the admiral, “we’ve got autonomous vehicles searching for mines, providing information through an unmanned surface vehicle to a UAV [flying drone] which is providing info to a command and control system.”
“What you’ve done by that,” he explained, “is reduced the number of people in the search element but you are able to place more people into the command, control and analysis element.”
The discussion came as part of the Navy’s ongoing Unmanned Warrior drone exercise taking place off the north-west coast of Scotland. Naval personnel, along with civilian firms, are putting 40 different types of drone through their paces, from traditional aerial surveillance craft to autonomous mini-submarines.
Commander Peter Pipkin, the RN’s fleet robotics officer, joined the admiral in giving examples of current naval thinking on uses for drones.
“Putting up unmanned search and reconnaissance aircraft... seems to be an obvious area where you could exploit these technologies,” he said.
Pipkin added that for a deployed ship or flotilla to physically examine areas of interest over the horizon means deploying an expensive helicopter and its crew into - potentially - harm’s way.
“The level of integration [for drones at present] is such that systems could be tasked to do that dull bit and understand that they need to flag up to the operator when their input is needed,” continued Cdr Pipkin.
In the civilian maritime world the use of drones is chiefly to replace expensive humans and ships in mundane tasks such as surveying the seabed and similar data-gathering exercises, but the naval use of drones is much more intensive than that. As well as observing that something unusual is going on, drones also need to be able to react to whatever that event is.
While El Reg was assured that current British policy is for a human always to be in the loop, particularly when armed drones such as the infamous Reapers are being used, the question of how much work a human can usefully do while supervising a number of drones is an intriguing one.
While at defence research agency Qinetiq’s Portsdown facility, we were shown a prototype one-man-for-many-drones control system. It was immediately apparent just how quickly operators' workloads escalated if one sea-surface drone encountered a situation requiring sustained human interaction - and how quickly that workload could become overwhelming if more than one drone needed his attention.
The challenge for today’s drone control system designers is to hone their systems into being able to make autonomous decisions at a certain level. Do you need a human to tell you to make a course change in order to avoid a collision with another vessel, or is that something which the drone can be programmed to do without further input? If you’ve just run into something unexpected - say, an uncharted rock - can a human do anything useful at that point other than pinpointing your location for a rescue crew to recover the drone?
The possibilities for machine learning in the development of tomorrow’s drone control systems is obvious. Whether industry is able to keep up with that challenge is another question. ®