Thursday, October 9, 2014

The Navy's Swarming Robot Boats: Sorting Through the Hype

By now, most readers are familiar with the U.S. Navy's new swarming drone experiment using the CARACaS (Control Architecture for Robotic Agent Command and Sensing)  system for automating small surface craft. The hype surrounding this development is significant, and some of it is rightfully deserved. Automated boats will find a place in future naval operations, but their capabilities and limitations must be more fully understood before that happens. 

August 2014 - A swarm of CARACaS equipped patrol boats on the James River - U.S. Navy image.
The primary benefits of autonomous unmanned vessels are longer endurance than manned patrol boats, and of course, a reduction in risk to human sailors.  Naval budgeteers consistently lament the cost of personnel, so ostensibly, automated boats will be less expensive than training and maintaining human crews.  On the other hand, the Navy's force protection boat crews, most of them resident in the Naval Expeditionary Combat Command are a trivial component of the overall budget, and many of them are part time reservists, which cost even less than their active duty counter-parts.

Perhaps the biggest advantage is that CARACaS is basically "plug and play" and platform agnostic. That means that the system can be rapidly forward deployed by air in a small package and fitted on whatever patrol boats are available versus having to transport a much heavier dedicated USV via expensive air or sealift. And since it reportedly only cost about $2,000 to outfit a patrol boat with CARACaS, outfitting several existing boats might be orders of magnitude cheaper than dedicated USVs. So creating a small boat swarm could be achieved fairly rapidly and cheaply if one were needed.

Myths & Limitations
The CARACaS boats have so far been employed from inland ports, limiting their utility to the fleet. It would be a non-trivial technical feat to launch such a large and recover a swarm of boats from a navy ship at sea.  Only a handful of ships in the fleet could realistically embark more than a couple of unmanned vessels. The U.S. Navy's cruisers and destroyers usually only carry two rigid hull inflatable boats, which are dedicated for other missions such as boarding operations.  Larger amphibious ships could carry several more boats in their well decks, but then would be unable to embark their usual load of landing craft.  That is not to say that several combatants couldn't launch their boats simultaneously and then the swarm would aggregate, but that is unlikely given how disbursed ships operate these days.  Of course, these autonomous boats could also be launched from a coastal base as they were during the experiment, but that would limit their employment to fairly close to shore.  Moreover, new software algorithms will need to be developed to help these boats safely approach their motherships, launch, and recover in any sort of sea states.

As with any unmanned system, options to fix mechanical malfunctions about CARACaS-equipped craft would be significantly limited. Minor engine problems or weapons jams that could be easily corrected by an onboard crewman will likely render an unmanned platform ineffective without some form of redundancy.

This technology is still immature and close-in maritime force protection as demonstrated in the James River involves high speeds and rapid decision making.  The ethical issues concerning lethal automated drones have been covered extensively elsewhere.  But even with experienced human boat operators, making a decision whether or not to engage a suspicious contact vessel is difficult. Generally, when threatened with a suspicious contact, a gradual escalation of force occurs that can take anywhere from a couple of minutes to a handful of seconds.  Warnings, shouldering, non-lethal, and lethal weapons, can be employed in various combinations to deal with a threat. In a fully unmanned or even autonomous system, communications latency, jamming, and operator inattentiveness could all play a role in this system failing to protect its escorted high value asset, or in creating an undesirable escalation of force.

Concerning the CARACaS swarm, Rear Admiral Matthew Klunder, Chief of Naval Research commented that "If [USS] Cole had been supported by autonomous unmanned surface vessels, they could have stopped that attack."  This statement could be theoretically, true, if USS Cole had been capable of carrying and deploying a number of USVs (it couldn't) and if no other counter-measures were available.  These defensive measures could have been as simple as enhanced rules of engagement, better intelligence on the threat at the time in Aden, one or more manned picket boats, or simply avoiding Yemen as a refueling stop all together.  

CARACaS patrol craft escort a simulated "high value" vessel near the Navy's Ghost Fleet - U.S. Navy Photo.
The future addition of “automated target recognition” systems will help the USVs discern the difference between friendly and potentially hostile craft.  Additional automation will also enable CARACaS-equipped boats to follow internationally-recognized rules of the role which will enable them to operate in more congested maritime environments.  This level of autonomy will be necessary before the system can operate in any sort of constrained waters, which is generally where force protection missions occur.

Aside from the force protection mission demonstrated by ONR, automated small boat swarms could support a number of other naval missions.  A better use of this technology than for force protection might be to enable autonomous swarms of mine hunting craft, such as the Mine Hunting Unmanned Surface Vehicle.  Mine hunting and other underwater searches, such as that performed for the missing MH370 jetliner, are more laborious and painstaking than coastal patrol and interdiction.  This sort of technology could enable them to be completed more efficiently and rapidly.  

Offensive unmanned boat swarms could harass, impede, and attack enemy shipping without risking the lives of friendly sailors.  Depending on the sea state, these swarms could remain dormant for some time in a maritime choke point, then activate as an enemy vessel began to transit through the area. They could also serve as forward observers to target enemy craft for destruction from airpower or ship-launched missiles. 

The collection of intelligence is another mission for which this technology might work well. Semisubmersible or indigenous autonomous craft are low visibility and could be outfitted with sensors and the CARACaS package.   For example, a small unmanned fishing vessel equipped with sensors could probably infiltrate an adversary's port of interest without causing any alarm much easier than a recognizable naval patrol boat.  USVs can even be equipped to carry UAVs to extend their sensor range, which is one of the options under consideration for the Littoral Combat Ship's mine-sweeping CUSV.

Bottom line, CARACaS is a worthwhile research effort, but probably not for the reasons the Navy currently envisions.  The demonstrations thus far have been impressive, but also highly scripted and controlled.  Future tactical employment of automated boat swarms in any meaningful manner is likely at least a decade away.  But the plug-and-play nature of singular CARACaS-fitted boats will probably happen much sooner.