Feedback Request for "The Operational Environment and the Changing Character of Future Warfare"

Cole,


Thanks for the comments. Computer-controlled and networked physical systems (cyber-physical systems) have been incrementally improving and it is reasonable to assume that they will continue to do so. I certainly agree with the NSF contention that CPS advances underway will change the manner in which humans will interact with machines. It is also reasonable to “think out of the box” concerning asymmetric warfare possibilities. However, in my opinion, it is not reasonable to assume outcomes for which there is no current path for achieving the assumptions. Planning that there will be sentient machines when there is no scientific basis for that assumption makes for good books and movies but will very likely not result in more capable combat units.


Probably the largest commercial manufacturing development activity underway today is the set of efforts by many (all?) of the car manufacturing companies to build autonomous driving vehicles. These efforts are underway partially due to the decades-long efforts at DARPA and other government agencies in many countries to create autonomous vehicle technology. TRADOC (Fort Benning combat developments) provided the requirements document back in 1988 which stated that the US Army needs an autonomous robotic capability for conducting intelligence, reconnaissance, and surveillance activities. Senator Warner of Virginia led efforts in the Senate to add target engagement capabilities to the set of existing requirements for autonomous robots. To date, autonomous robotic vehicles which provide ISR and target engagement capabilities to units remain largely a research effort. What has worked well for many tears is a direct extension of what worked during WWII using radio technology to enable humans to drive remote vehicles. Swarm technology has made dramatic advances recently to achieve patterns of air, ground, and sea movement. Swarm technology currently has no path for meeting ISR and target engagement requirements. However, success of the car manufactures autonomous highway vehicles will mean success of the ongoing efforts of the logistics community to have a group of autonomous resupply trucks driving down the road or autonomous pack mules following a squad through the woods. This will not be sufficient to enable lethal robots to join combat units and follow variable rules of engagement (ROE).


In fact, a persistent failure of robotic technologies has been a failure to enable machine understanding of human intent to be "close enough" to human understanding of human intent. The planning process for creation and dissemination of operations orders enables humans to understand the intent of the commander for a given operation. Humans adapt the plan "on the fly" to meet the intent of the commander. We have no technology or process (a TTP) for enabling machines to understand command intent for a given operation. To my knowledge, there is currently no approach, or a program for developing such an approach, which will enable machines to achieve that outcome. Thus, whether the operation is completing the set of Tasks, Conditions, and Standards for a mechanic to change a tire, or whether the operation is for one rifle squad to provide covering fire for another rifle squad during a tactical movement, there is no reasonable expectation that robots will function well as part of a combat team. That does not mean that we should not continue to investigate the possibility of making a revolutionary breakthrough in building sentient machines. However, for me it does mean that we should not plan for such an outcome. What does make a great deal of sense to me is to create scenarios of future warfighting outcomes which exploit the ongoing revolution in information system technologies, cyber physical system technologies, and the possible emergence of the "Age of Biology." Each of these possibilities are discussed in the paper.

I do agree with your comments concerning the impacts on future outcomes of resource allocations by ourselves and by other nations. Certainly for most of our history we had a very small Army, Navy and Marine Corps between the various wars. After WWI the Army retained the largest force in our history but it was still less than 100,000 soldiers. Even after WWII we cut back the force structure to be very small compared to the millions who served during WWII. Only since the Korean War and the emergence of weapons of mass destruction have we retained a large standing force structure and the associated capability for building weapon systems which incrementally improve combat capabilities of that force structure. Building a future force while retaining sufficient capability for meeting current national security commitments will most probably continue to be a high risk endeavor.

Cheers,
John

John,

For a robot to be sentient, there is an implication of multiple sensors and computing power not likely to be cheap. I just read about a small UAS called Coyote in the realm of $15k a piece which is probably before sensors, warheads, and AI computing power for more than organizing and deconflicting the swarm (probably not including deconfliction with manned aircraft!). Looking at systems like LOCUST also leads me to speculate endurance of no more than half an hour if that. Assuming 30-100 UAS as suggested in articles are part of the swarm, at $15k each that is 450k to 1.5 million dollars. It is pretty easy to see all that money being wasted being unable to find a target in the short time available, and a potential hazard to ground personnel and civilians when the swarm runs out of gas. Then there is that airspace hazard.

If launched from three F/A-18 as in one published experiment, it implies a significant standoff range to keep the manned aircraft safe from air defenses. It also means that something like several JDAM or multiple gliding small diameter bombs could be more cost and lethally effective. Plus, half the swarms 30 minute endurance could be wasted just getting to the target area kill box where the enemy has moved or sought cover. Perhaps the enemy force has lasers or radio frequency/cyber jammers or microwave weapons as you quickly read about through brief searches.

Yes, self-driving cars have a commercial future sharing the road with driven vehicles. Likewise, unmanned systems must share the air, ground, and sea spaces with manned systems, and stay close enough to manned systems for control data links with ground, air, and sea unmanned systems. Believe me, I'm a true believer as evidenced by a manned-unmanned teaming article I wrote for Aviation Digest. My current job involves unmanned systems and it is quite humbling and illustrative that these systems work best NOT unmanned but with very capable manned direction from afar. I also worked FCS so understand the enormous potential, shortcomings, and costs of integrating manned and unmanned systems. Some of the crazy ideas they had for computerized mission command systems for OPORDs, etc, get to your points. But lethally and for information collection, I can see the former loader of a tank and cavalry vehicle controlling an unmanned ground vehicle and serving as a back-up to the autoloader. Systems like Switchblade show the potential of shorter-range lethal UAS. The idea of launching swarms from survivability equipment tubes on manned aircraft could have great merit.

However, there is a reason why we are building F-35s, have F-22s, and want a new bomber. There is a reason why manned helicopters must fly low and slow in some combat environments while higher flying UAS maintain stand-off and launch shorter range weapons. We hope to achieve air superiority rapidly so the threat of manned aircraft and air defense shoot-down of UAS is a reduced threat. We want to penetrate enemy defenses early in many conflicts delivering munitions initially in a stealthy manned aircraft configuration. That would "shoot down" the idea of 3 F/A-18s with a large signature pod on board that instead could be one or more gliding bombs launched from afar.

John,

For a robot to be sentient, there is an implication of multiple sensors and computing power not likely to be cheap. I just read about a small UAS called Coyote in the realm of $15k a piece which is probably before sensors, warheads, and AI computing power for more than organizing and deconflicting the swarm (probably not including deconfliction with manned aircraft!). Looking at systems like LOCUST also leads me to speculate endurance of no more than half an hour if that. Assuming 30-100 UAS as suggested in articles are part of the swarm, at $15k each that is 450k to 1.5 million dollars. It is pretty easy to see all that money being wasted being unable to find a target in the short time available, and a potential hazard to ground personnel and civilians when the swarm runs out of gas. Then there is that airspace hazard.

If launched from three F/A-18 as in one published experiment, it implies a significant standoff range to keep the manned aircraft safe from air defenses. It also means that something like several JDAM or multiple gliding small diameter bombs could be more cost and lethally effective. Plus, half the swarms 30 minute endurance could be wasted just getting to the target area kill box where the enemy has moved or sought cover. Perhaps the enemy force has lasers or radio frequency/cyber jammers or microwave weapons as you quickly read about through brief searches.

Yes, self-driving cars have a commercial future sharing the road with driven vehicles. Likewise, unmanned systems must share the air, ground, and sea spaces with manned systems, and stay close enough to manned systems for control data links with ground, air, and sea unmanned systems. Believe me, I'm a true believer as evidenced by a manned-unmanned teaming article I wrote for Aviation Digest. My current job involves unmanned systems and it is quite humbling and illustrative that these systems work best NOT unmanned but with very capable manned direction from afar. I also worked FCS so understand the enormous potential, shortcomings, and costs of integrating manned and unmanned systems. Some of the crazy ideas they had for computerized mission command systems for OPORDs, etc, get to your points. But lethally and for information collection, I can see the former loader of a tank and cavalry vehicle controlling an unmanned ground vehicle and serving as a back-up to the autoloader. Systems like Switchblade show the potential of shorter-range lethal UAS. The idea of launching swarms from survivability equipment tubes on manned aircraft could have great merit.

However, there is a reason why we are building F-35s, have F-22s, and want a new bomber. There is a reason why manned helicopters must fly low and slow in some combat environments while higher flying UAS maintain stand-off and launch shorter range weapons. We hope to achieve air superiority rapidly so the threat of manned aircraft and air defense shoot-down of UAS is a reduced threat. We want to penetrate enemy defenses early in many conflicts delivering munitions initially in a stealthy manned aircraft configuration. That would "shoot down" the idea of 3 F/A-18s with a large signature pod on board that instead could be one or more gliding bombs launched from afar.