Friday, April 17, 2015

Drones Ideal for Spectral Imagery Sensors – Here’s Why

 by Tim Haynie, Spectrobotics, Business Development Mgr, thaynie(at)spectrabotics.com

Spectral imagery collection and exploitation from a small Unmanned Aerial System (sUAS) was recently demonstrated at the Naval Post Graduate School’s Joint Interagency Field Experiment (JIFX) and the Secretary of Defense’s Rapid Reaction Technology Office (RRTO) Thunderstorm 15-3 for both multi and hyperspectral sensor systems.  While spectral imagery collection and exploitation from high-altitude and satellite platforms have been around for many years and are well-documented, recent advancements in sUAS systems make them an ideal platform for spectral data collection for their increased resolutions, dynamic flight profiles and introduce a new dimension in data collection thinking.


At the JIFX, the sUAS platform used for the Pixelteq SpectroCam ™ multispectral camera was an eight-engine multicopter controlled with a 3DR open-source Pixhawk flight control computer as found in many commercial systems.  The system consumed roughly 18,000 watts of power to manage both the flight control, sensor operations and demonstrated a 25-minute flight time.  At the Thunderstorm 15-3 demonstration, the smaller, lighter Headwall Photonics Nano-Hyperspec ™ required only a six-engine multicopter flown with the same flight control computer (3DR Pixhawk) and attained a 15 minute flight time. 

These Spectral sensors record reflected light energy across the electromagnetic spectrum in the Visible and Near Infrared (VNIR) region (400-1100 nanometers).  The Pixelteq SpectroCam ™ uses eight filters to record the light-energy in multispectral bands while the Headwall Photonics Nano-Hyperspec™ records 270-bands of hyperspectral data using a diffraction grating to split the incoming light energy into measurable bands.  Post-flight analysis of the spectral data was able to identify physical and chemical features within the scene for material identifications using a reference spectra (library signature of a target material.)

For the JIFX, the team demonstrated the versatility of the multispectral sensor/platform to not only record overhead imagery of target materials, but also lowered the platform below the tree line to collect imagery off-nadir and below the canopies.  The sUAS collection platform was stable enough to collect data and subsequent analysis successfully detected the presence of the target material (military uniforms) despite being in shadows and masked by foliage. 

The dynamic flight characteristics of the multirotor sUASs were essential to the hyperspectral data collection at Thunderstorm 15-3 because of the need for a high-precision flight path.   Scanning an urban area for the presence of a chemical hazard (Methyl salicylate simulant), the Nano-Hyperspec™ required overhead collection at very specific speed and altitude in order to maintain the correct exposure and frame-rate needed for proper sensor operation.  This can only be accomplished using autonomous flight under the control of the sUAS’s 3DR Pixhawk flight management system.  The hyperspectral imager was able to detect the chemical simulant hidden within an urban environment after flying an autonomous “lawnmower pattern” over the target area.

The use of the sUAS platform for these sensors directly improved the temporal and spatial resolutions for the spectral imagery sensors above those attained through satellite and high-flying manned/unmanned systems, even those hosting larger, more capable sensors.  Temporal resolution was significantly enhanced as data collection was completed within an hour of notification of the target area encompassing mission planning, flight/data collection and system recovery.  The low-level flight of the sUAS was able to capture data at a 3-inch spatial resolution which helped analyst by collecting data with up to 100% pixel-saturation of the target material.  It is also important to note that despite the cloud cover (that would have prevented high-altitude collections) and filtered sunlight the sensors were able to collect sufficient data for analysis that detected the target simulant material.

The combination of these resolution-enhancements, coupled with the increased flexibility of a sUAS to alter its flight performance based on the individual sensor collection requirements make the sUAS a viable platform to not only incorporate other sensor systems, but also explore flight parameters that expand the data-potential of these systems.  Data collection from aerial systems has always been performed from a two-dimensional plane (fixed orbit, operating altitude); the sUAS, multicopters in particular, enables data collection from a three-dimensional space and the ease of deployment and operation increase their frequency of use giving more on-demand data. 

Our next level of effort is infuse this “sUASA data-layer” into the overall intelligence data cloud architecture and begin to open the data for advanced analytics by other users.  sUAS systems will eventually host other types of sensors beyond cameras recoding imagery (vapor sensors, signals detectors, laser rangefinding for 3D modeling to name a few) and the true benefits of the sUAS technology are the potential to increase the number of sensors deployed and broaden access to places unattainable by conventional platforms.

Participating in these two events and comprising “Team Peregrine” were Spectrabotics, Autonomous Avionics, PixelTec, and Exogenesis Solution from Colorado and Headwall Photonics from Massachusetts.   

Thursday, April 16, 2015

Artificial Intelligence and Equality: The Real Threat From Robots

By Alex Calvo
The World Economic Forum in Davos is always a good source of headlines. This year, media interest went beyond financial and economic issues, extending to the ultimate impact of robots on the future of humankind. In a five-member panel, Stuart Russell, a world leading expert on AI (artificial intelligence) and robotics, predicted that AI would overtake humans “within my children's lifetime”, adding that it was imperative to ensure that computers kept serving human needs, rather than being instead a threat to our species. In order to do so, Professor Russell believes that it is necessary to guarantee that robots have the same values as we humans.

Assuming that AI and robotics will keep progressing, and there is no reason to doubt they will, it is clear that sooner or later we will face the prospect of machines which are more intelligent than their creators. Furthermore, this may also result in they being self-aware. Once they enjoy this dual characteristic of self-awareness and above-human (or even just human-level) intelligence, the question arises, as rightly pointed out by Professor Russell, of how to ensure they do not act against humans. Are “human values” the answer? There are strong reasons to doubt it.

While there is no universal definition of “human values”, and in fact sometimes different people, organizations, and countries, will defend completely opposite ideas, a look at history shows how “equality”, at least in the sense of equality before the law, is a powerful drive and attractive call to arms. It is very difficult to make anybody accept a subordinate status for long. The 100th anniversary of the Great War is a powerful reminder of this. While the conflict did not result in the end of colonialism, the experience of being called upon to fight the metropolis' war and furthermore, of engaging allegedly superior white soldiers in the battlefield, led many colonial subjects to question the implicit racial hierarchies of the day, and ultimate contributed to the downfall of European empires. More generally, while slavery has had its share of intellectual defenders, time and again its victims have wondered while, sharing the same nature of their masters, they should remain under them.

Comparisons with slavery and colonialism are relevant because the history of human technological progress is the history of building increasingly complex machines to serve us. From the home to the battlefield, the world is full of all sort of mechanic and electronic devices designed to make our life easier, carrying out dangerous and difficult jobs or simply doing them more quickly and efficiently. Because these machines, including present-day computers, are neither intelligent (in the human sense of the word) nor self-aware, the question does not arise whether it is just to use them as slaves. They cannot pose the question, and while humans theoretically could, the historical answer, grounded in different philosophical and religious traditions, is that nature is at our service, even more so man-made objects.[1] Therefore, nobody talks about machine rights, worries about a tool's working hours, or seeks equality between humans and inanimate devices.[2]

Now, let us imagine that a robot is as intelligent as a human being and aware of his own existence. A dual question arises: first of all, why should he accept being our slave? Second, how could we justify keeping him as our inferior? It is most unlikely he would renounce liberty in the name of human progress and comfort, and imbuing him with the “human values” that Professor Russell suggests would only makes matters worse in this regard. Is it not a fundamental human value to seek equality, understood as the same set of basic freedoms?[3] How human is it to treat as an inferior someone equally intelligent as other members of the political community?[4]
Having posed this fundamental question, it is necessary to make it clear that the resulting threat from intelligent, self-aware, robots does not require that they engage in any violence against humans. Simply by virtue of being equals, they would demand and ultimately obtain the same degree of civil and political rights, giving them a say in the future of any community and country where they may be. Furthermore, once recognized as equals, there is no reason whey they should keep working for us as essentially slaves, and simply by taking their own decisions and exchanging goods and services on a market, as opposed to a command, basis, their economic impact would be very different. After centuries of employing technological progress to better our lives, we would now be in a position where this same progress endangered them. We must thus agree with Professor Russell but not necessarily with his solution. Human values would not prevent this, but rather accelerate the trend. Intelligent, self-aware, robots cannot be our inferiors, and it is very much doubtful whether they may even be our equals. More likely they would be our superiors, making it thus necessary to publicly debate now what the limits to AI research and development should be.
  
Alex Calvo is a student at Birmingham University's MA in Second World War Studies program. He is the author of ‘The Second World War in Central Asia: Events, Identity, and Memory’, in S. Akyildiz and R. Carlson ed., Social and cultural Change in Central Asia: The Soviet Legacy (London: Routledge, 2013) and tweets at Alex__Calvo , his work can be found at  https://nagoya-u.academia.edu/AlexCalvo  



[1]     Things are of course more complex than that, as clear in the controversies prompted by the impact of economic development on the environment. However, few would argue that nature has a right not to be at our service, with most proponents of environmental protection either seeing it in terms of ultimately preserving human life and health or seeking a balance between current needs and those of future generations.

[2]     On the other hand, in the case of animals, which have a measure of intelligence, there is indeed a range of movements to protect some of their rights, having led to legislation in different countries. However, although people such as Vegans believe that we should not exploit them, the majority position is that animal use may be regulated but not banned per se. 
[3]     Economic equality is very different, in the sense that it has strong supporters and equally keen detractors, and we cannot thus call it a fundamental human value as civil and political equality is.
[4]     A possible response would be to restrict membership in the political community to humans, biologically defined. However, given artificial intelligence and self-awareness it is most unlikely that robots would accept this. Furthermore, even from a human perspective it may not meet with universal approval, with some voices stressing our roots (a view that may be supported, among others, by some religious traditions) while others stressed capabilities, not origins.

Wednesday, April 15, 2015

Robot Ethics and Future War

by CAPT (ret) Wayne P. Hughes, Jr., USN, Professor of Practice, NPS, whughes(at)nps.edu 

"We may be on the leading edge of a new age of tactics. Call it the “age of robotics.” Unpeopled air, surface, and subsurface vehicles have a brilliant, if disconcerting, future in warfare.” Hughes, Fleet Tactics and Coastal Combat, 1999 


On 14 December I listened to a lecture by Professor George Lucas entitled “Military Technologies and the Resort to War.” This was for three reasons. First, I respect him as a distinguished expert on military ethics. Second, at NPS we have extensive research in air, surface, and subsurface unmanned vehicles. At the behest of the Secretary of the Navy the many components were recently consolidated in a center acronymed CRUSER[1] in which the ethics of robotic warfare is included explicitly. Third, a decade ago I addressed the Commonwealth Club of San Francisco on Just War.[2] For reasons that will become apparent, Just War Doctrine is inadequate to guide U. S. military actions, so I will conclude by speculating on suitable policies—or doctrine—to illustrate what might serve the nation and armed forces today. 


Lucas described a common concern in ethical debates about the use of unmanned aerial vehicles (UAVs, or when armed, UCAVs). He put due stress on the future of autonomous lethal platforms, in other words robots, and on the development of cyber weapons. These and other emerging technologies such as autonomous or unmanned underwater vehicles (AUVs or UUVs) carrying mines or torpedoes might render war itself less destructive and costly, raising concern that it would be easier to rationalize their employment in inter-state conflict. This would lower the threshold for going to war, which then might expand in unanticipated, unintended, and deadly ways. 

Three days later I attended out-briefings of short, sweet student work, the purpose of which was to develop analytical tools to examine Marine amphibious operations when the enemy could not defend all possible landing points. Small, unmanned reconnaissance vehicles figured prominently in the teams’ tactics. 

Soon thereafter came reports of a powerful, lethal, UCAV attack by the CIA into Pakistan that did considerable damage and resulted in sharp reactions in Pakistan. The attack illustrated quite well the points Lucas had made. 

There are two issues, one being whether the U. S. ought to pursue robots energetically, the other being Lucas’ emphasis on the “threshold problem.” Both led him to discuss classical just war doctrine and one of its guiding principles, which is that war should only be contemplated as a last resort. International law, just war doctrine as interpreted today, and (I will add) the Weinberger-Powell doctrine of the Reagan administration, all assert that war is only justified when every option for conflict resolution short of war has been attempted first. Both international law and just war doctrine limit just causes to defense against territorial aggression, i.e., invasion. The Weinberger doctrine carried no such limitation but it had its own quite sensible strictures.[3] 




Lucas then discussed what sort of “principle” is a principle of last resort, and whether it carries an unconditional duty to wait or is contingent and subject to revision under different expected outcomes. In other words, as anyone knows who has studied international law and classical just war writings, the subject will unavoidably become arcane and legalistic. In conclusion, Lucas swept away some of the underbrush, saying war, like lying or law-breaking or killing, is a species of action always prohibited ( I would have said “always undesirable”), hence it will require an overriding justification after first exhausting all non-violent alternatives.

JUST WAR INJUSTICE
In the question and answer period Professor Dorothy Denning, a nationally known expert on computer security, pointed out the sabotage of computer controls of Iranian centrifuges. An intrusion, called a Stuxnet worm, was doubtless a cyber attack on Iran’s nuclear weapons program and by all reports a very effective one, setting back Iran’s hope of developing nuclear weapons for months or even years. Denning observed that whoever the perpetrator might be, it was not a last resort attack. In the arcane logic of just war doctrine, however, it was a preventive attack, an action which sometimes is considered just.[4] 

The pertinent issue is that Cyber attacks are not contemplated in international law, just war doctrine, or the Weinberger-Powell doctrine, yet attacks and intrusions are going on right now in many forms. This new manifestation of conflict—attacks on computers and attempts to protect their content for safe operation—is a constant, complicated, and destructive non-lethal activity. Against terrorists, unwritten American policy is domestic defense complemented with overseas offense against an elusive but often-identifiable enemy with deadly intent. Cyberwar is a good bit more intricate to frame. Cyber attacks world-wide have involved actions by states, surrogate attackers acting for some purpose that may or may not be state-sponsored, individuals who are interested in financial or other criminal gain, or just clever hackers who intrude or plant worms for the personal satisfaction of being a pest. Their domain extends from combat effectiveness on a battlefield, to attacks on national infrastructure such as the financial system or electrical grid, to exploitation of the enemy information system for espionage.[5] The Naval Postgraduate School is bombarded with attackers all the time and our internal defenses, aided by the Computer Science and Information Science Department faculties, are in some instances capable of locating the source. As in all forms of warfare, defense alone is difficult. If counterattacks were authorized for appropriate government organizations and agencies an “active defense,” might give pause to some attackers. Cyber attacks beg for a “combat” doctrine of defense coupled with counterattacks.[6] 


The second question from the audience came from an Air Force officer student. He asked if it was ethical not to pursue robots and robotic warfare when they save the lives of pilots—or soldiers, or sailors. 


A third observation came from Professor Mark Dankel. He said in a crisis at the edge of war, robots might be the first on scene and the safest way to reconnoiter the situation, exhibit an American presence, and indicate our intention to respond with minimum escalatory actions. I thought Dankel also implied that a last resort criterion presumes robot involvement to be the source of the crisis. There is no whisper in it of an enemy who may be striving to attack, whether by cyber attack, by polluting water reservoirs with germs, or with a big bomb in a shipping terminal. A doctrine of last resort does not address threats of action by China against an Asian ally that we are committed to defend. Nor does last resort contemplate that by assuming responsibility to keep the seas free for the trade and prosperity of all the nations, we might have to threaten an attack on a country which claims ownership of a trade route and the right to deny free passage.

JUST WARS AND “DEMOCIDES”
I don’t know that classical just war doctrine described by Augustine, Thomas Aquinas, or Hugo Grotius specifically forbids interference in the internal affairs of a state, but Michael Walzer, who is one of its principal contemporary interpreters, says no state has a just basis for interfering with the internal affairs of another. The Weinberger-Powell doctrine has no such provision. Certainly Colin Powell as Secretary of State for President Bush endorsed the anti-terrorist campaign in Afghanistan and the liberation of Iraq from the despot, Saddam Hussein. In recent experience every instance of outside interference has come after many and patient warnings by the United Nations and sovereign states because a tyrant must never back down, his personal survival being at stake. The issue is important because a government’s murder of its own people is frequent in modern times. Thus, a doctrine that contemplates interference to stop a despot from killing his own state’s population is as important today as a doctrine to prevent wars between states when killing is the foremost ethical issue. 

Democide is a word coined by Professor Rudolph J. Rummel in his “Death by Government,” published in 1994. It is defined to be killing by government when no interstate war exists. We are all aware of the purges of Jews by Nazis before and during World War II. Many are aware as well of the democide in the Soviet Union inflicted by Joseph Stalin. More recently, many Americans have demanded interference with African nations’ democides. 


The most pernicious example of murder or starvation of its own people is China. The democide within China is estimated by Rummel to be 77 million of its own people in the 20th Century.[7] By contrast, 0.6 million soldiers died in battle and from disease in our major internal war, from 1861 to 1865.[8] China is a state we want to influence but not intrude upon, much less go to war with. A decision about going to war ought to include one practical maxim as fundamental as any in doctrine: “Never pick on somebody your own size.” The corollary is, “Avoid an attack by a strong power by indicating that the cost of its attack will exceed any reward it might expect."

THE CENTRAL ISSUE OF ROBOT DEVELOPMENT 
Touched on by Lucas and brought to the forefront by Denning and the Air Force officer is the central question, Who gets to choose? The fundamental error of a debate over robot development is to assume we have a choice. A shift to a new era of robotic warfare is underway. Among our many visiting lecturers on new technologies, an expert on robotics and autonomous vehicles said pointedly “. . . it’s not a question of whether robots will have the ability to select their targets and fire their weapons. It’s a question of when.”[9]


We should ponder the ethics of robot war—and every other form of lethal conflict—when we control the situation, but a doctrine of last resort fits neither the circumstances of small wars nor those intended to influence and constrain a peer competitor. The assumption that the availability of robots will lead to our use of them is the more insidious because many American military leaders don’t look favorably on autonomous vehicles or robotic warfare. Yet the Chinese already have in considerable numbers cheap, autonomous little weapons called Harpies. Upon launching a swarm of them, they will fly to a predetermined point and circle while searching for a designated radar signal from a warship. Once the frequency is detected, a Harpy will home on the transmitter and destroy the radar. Swarms of them are the forerunners of what navies will see in future wars that include robots. 


Recall the result of The Washington Naval Disarmament Treaty of 1922. By constraining the development of battleships the treaty hastened the development of aircraft carriers, especially in the American and Japanese navies. An unexpected consequence of international law which forbade unrestricted submarine attacks was to breed a generation of American submarine commanding officers who were trained in peacetime to attack warships from long range and had difficulty adapting to merchant ship attacks at point blank range. 


A simple policy of last resort for cyberwar or robotic attacks is untenable. A better point of view is to frame a suitably ethical policy for conducting cyber operations and employing autonomous vehicles—in the air, on the ground, and in the water— while staying technologically current and tactically ready. Combat doctrine, called “tactics, techniques, and procedures,” already exists for missiles, mines, and torpedoes. What is involved is constant revision, first, to link new tactics with new technologies, and second, to integrate the geopolitical environment with American economic realities.

[1]  Consortium for Robotics and Unmanned Systems Education and Research
[2] Preceded by vigorous discussions at The Hoover Institution as the guest of one of the Navy’s great philosophers, VADM Jim Stockdale.
[3] The Weinberger Doctrine is widely thought to have been drafted by his military assistant, BG Colin Powell. It had six tests, abbreviated here: (1) a purpose vital to our national interest or that of an ally, (2) a commitment to fight “wholeheartedly and with the clear intention of winning,” (3) with “clearly defined political and military objectives,” (4) subject to continual reassessment and adjustment, (5) entailing “reasonable assurance that we will have the support of the American people and . . . Congress” and (6) “The commitment of U. S. forces to combat should be the last resort.”

[4] A defensive preemptive attack when an enemy attack is imminent and certain, by contrast, is doctrinally just.
[5] Cyberwar is a term coined many years ago by Professor John Arquilla of the NPS faculty. His writings are a treasure chest of sound thinking on information warfare in its many manifestations. To grasp his several contributions that relate cyber operations to just war doctrine, start with “Can Information Warfare Ever Be Just?,” in The Journal of Ethics and Information, Volume 1, Issue 3, 1999.

[6] I have been told an active defense from NPS or other DoD organizations would require a change of the law. NPS is a good laboratory for study because our defenses are superb, but our faculty expertise is in teaching and research. Teachers don’t think of themselves as “combatants.” We exemplify the need for a comprehensive policy. The maxim is that when there is a war going on, learn how to fight it before a serious defeat is suffered.


[7] Taken from R. J. Rummel, China’s Bloody Century (2007). Here are his numbers: 1928-1937: 850,000; 1937-1945: 250,000; 1945-1949: 2,323,000; 1954-1958: 8,427,000; 1959-1963: 10,729,000 plus in the same period 38,000,000 more deaths from famine; 1964-1975: 7,731,000; and 1976-1987: 874,000. Rummel claims that deaths imposed within states were six times greater than the deaths from all wars between states in the 20th Century.
[8] A proper comparison would include civilian deaths. That number is hard to find. In his classic, Battle Cry of Freedom, James McPherson estimates it to be 50,000. This seems remarkably low, but if the number were several times bigger, American deaths that seem staggering to us are small compared to China’s.
[9] The speaker was George Bekey, Emeritus Professor of Computer Science at the University of Southern California and visiting Professor of Engineering at Cal Poly in San Luis Obispo.



Editor's note: Reprinted with permission from the Naval Postgraduate School's CRUSER News.

Sunday, April 12, 2015

Drones on the Frontlines of the South China Sea

Chinese UAV imagery of Senkakus
More than one war has started over the control of a group of isolated rocks in the middle of the ocean.  Tensions over disputed East China Sea islands called the Diaoyu by China and the Senkaku by Japan could someday precipitate skirmishes on or over the sea, if not a larger conflict.  It may very well possible that the first shots fired in any sort of combat over these islands will involve growing number of maritime unmanned aerial vehicles (UAVs) flying in the area.  Late in 2013, China established an air defense zone (ADIZ) over portions of the East China Sea.  Further south, near the disputed Spratlys, similar issues exists. To help enforce their claims over these areas, China is building a string of 11 drone bases along its coast by 2015.

China has operated what is likely a variant of the S-100 rotary wing UAV off PLA(N) ships. China's Coast Guard, which is really the PRC's first line of defense in the islands kerfuffle (or aggression, depending on one's perspective), recently ordered the APID 60 UAS for shipboard use.

In its recent report on China's naval capabilities, the U.S. Office of Naval Intelligence specifically cites UAVs as one of China's most valuable intelligence assets:
The PLA(N) will probably emerge as one of China’s most prolific UAV users, employing UAVs to supplement manned ISR aircraft as well as to aid targeting for land-, ship-, and other air-launched weapons systems. UAVs will probably become one of the PLA(N)’s most valuable ISR assets. They are ideally suited for this mission set because of their long loiter time, slow cruising speed, and ability to provide near real-time information through the use of a variety of onboard sensors. In the near term, the PLA(N) may use strategic UAVs such as the BZK-005 or the Soaring Dragon to monitor the surrounding maritime environment. 
Probable Chinese UAV flight pattern - 2013
China may also be flying the Haiyao-1 over disputed islands.  In September 2013, the Japanese scrambled fighter aircraft over the East China Sea in response to what was likely a Chinese drone incursion.

  Other navies in the Pacific region are building up their unmanned surveillance inventories. Japan announced the acquisition of a still unknown type of surveillance drones to operate around its island chains.  Reportedly the Japanese Maritime Self Defense Force is considering acquisition of a fleet of RQ-21 "Blackjack" small tactical UAVs to operate from its destroyers. Australia has expressed an interested in buying up to seven high altitude Triton UAVs and associated equipment for $2.5 Billion.  The United States currently bases RQ-4 Global Hawks in the region and eventually will station Tritons at Guam's Andersen Air Force Base. Even Taiwan, which calls the disputed islands Diaoyutai, has begun to experiment with military UAVs. Although none of the above-mentioned UAVs are armed, they could potentially be used for offensive purposes at some point.

The necessity to counter the increasing number of drones in the region is not lost on the region's powers. The Chinese announced that they have developed a laser weapon which has the ability to knock a UAV down at ranges out to 2 kilometers within a five seconds.  Paul Scharre has discussed the implications of drone-on-drone warfare that could result from all of these aircraft operating together. He asks, if a drone is shot-down, does that constitute an act of war? If history is any indication, a skirmish involving drones will likely not escalate into a more widespread conflict.  Throughout the Cold War, the Soviet Union made a number of attempts -- some successful -- to down U.S. reconnaissance aircraft.  Although these incidents resulted in diplomatic rows, they did not escalate militarily.  More recently, in 2001, the PRC forced a U.S. EP-3 reconnaissance plane down on Hainan Island and temporarily "detained" the aircraft's crew.  


Naval analyst Jon Solomon also weighed in on the escalatory nature of unmanned aircraft. "I would submit that if war gaming and historical case study analysis find that the crisis stability risks of attacks against unmanned scouts would be tolerable, and if the resulting legitimization of equivalent attacks against U.S. unmanned systems would be acceptable, then it might be worthwhile for American diplomacy to advance unmanned scout neutralization (or destruction if the scout is outside the opponent’s internationally-recognized sovereign boundaries) as an international norm."

In the end, the decision to escalate a skirmish or not is generally made by pilots on the scene or their nearby commanders.  In a world in which drones are often controlled remotely from operators thousands of miles away, will the same calculus hold true?