On Thursday, January 21, 2016, Northrop Grumman Chairman, Chief Executive Officer and President Wes Bush addressed the Wings Club Foundation Luncheon in New York City. Below are his remarks.
The Future of Autonomy
I’m delighted to be with you today.
Like me, you all understand and appreciate the central role of aerospace to our economy and our world. And this institution has been advancing that role for decades.
The Wings Club has provided an invaluable service to aerospace and our nation, and I feel honored to be among those who have addressed your membership over the years.
It certainly is an exciting time for aerospace in all its categories: Space, commercial aviation, military aviation, and all the many technologies that promise generational leaps in capability, affordability, safety, and return on investment.
Today, I’d like to focus on one such technology – specifically, unmanned aircraft.
Northrop Grumman focuses a great deal of our research and development, design and manufacturing efforts on this technology.
Not only on the unmanned platforms, but the multiple systems and technologies that enable unmanned aircraft to fulfill many missions for our nation, in both defense and scientific applications.
And, consistent with the kinds of unmanned aircraft we excel at, I’d like to increase the magnification a bit to focus on autonomous systems – aircraft that are truly robotic versus those that are remotely controlled by a human pilot. Because this is an aspect of unmanned aircraft that will change the game for both military and commercial aviation.
And let’s sharpen the focus even more to talk about the category of technology now under development that will provide the tipping point from which autonomous aircraft – be they military or commercial – will truly realize their potential.
I’m talking about cognitive autonomous systems. These are robotic systems that operate using the same actions we would expect from the judgment and ultimately, the ethics of a pilot.
As I’ll explain in a minute, it is not as mysterious as it sounds, but it is an enormous challenge to overcome. And autonomous aircraft will not grow into their full potential until we overcome it.
Let me give you a sense of how big that challenge is.
As you know, there are many companies here in the U.S. and around the world that are working to develop autonomous cars. The general public might assume that it’s not that tough a problem. GPS, combined with the right sensors and control laws, will keep the car on course, and will ensure that it doesn’t hit the car in front of it.
But let’s say a child runs out in front of that car. And let’s say that car is moving too fast to stop in time.
Now the car needs to figure out which direction to swerve to avoid the child.
And let’s say that if the car swerves to the left it risks hitting people on the sidewalk. But if it swerves to the right, it risks hitting on-coming traffic.
The systems controlling that car need to prioritize the risks, evaluate the potential harm of each option, and act on those evaluations. They need to do it instantly, and the results must be at least as good every time as one would expect from the best human driver.
And though we often reduce difficult concepts to human terms for ease of understanding, we need to be careful with comparisons between technology and the human brain. We don’t yet understand the human brain in all its complexity, but we can marvel at how it works.
However you want to think about it, the actions of an autonomous car must reflect the same concern for human life as a human driver. And ideally, such an autonomous vehicle would be able to act without a human’s judgment lapses or execution inadequacies.
That’s a much bigger problem than can be solved by a GPS receiver and a set of sensors.
Now, let’s apply that analogy to military aircraft. This will require a significant scaling-up of both the challenges and the solutions. The Navy, to its credit, has started to address the challenges of autonomy with its X-47B Unmanned Combat Air System.
Like the autonomous car, the cognitive systems on the Navy’s future autonomous aircraft ultimately will need to operate with judgment and reliability far better than the best conceivable human pilot.
But unlike the car, the aircraft must perform in three dimensions;
- At altitudes of tens of thousands of feet;
- At velocities of hundreds of miles per hour;
- In highly variable and adverse weather conditions, both in the air and maneuvering on the carrier deck;
- in hostile environments where the enemy is trying to shoot it down, or jam it, or hack it;
- Perhaps delivering lethal payloads onto targets — In which case, of course, for the foreseeable future, a human would always be in the loop;
- It must perform in confused environments in which some elements might be hostile and others friendly; and
- It must be able to do those things thousands of miles away.
And one last wrinkle: This aircraft has to do it all without a tail. Because if it has a tail, it won’t be stealthy. This is a pretty high bar for an autonomous strike aircraft.
But that bar is being met, step by step. In 2013, the X-47B prototype was successfully launched from, and recovered aboard, an aircraft carrier. To pass those tests, the aircraft had to maneuver on the deck in very close proximity to other aircraft and, more importantly, flight deck crew.
On approach to landing, it had to compensate for two bodies moving in three dimensions – itself and the pitching carrier deck – to touch down for a perfect trap.
I ask you to imagine the Navy’s accomplishment: 22 tons of unmanned, autonomous aircraft and fuel touching down on a pitching carrier deck in very close proximity to sailors and other aircraft. And again – without a tail.
For that accomplishment, the Navy’s X-47B team won one of aviation’s premier prizes, the Collier trophy.
But that was just a first step.
The next step focused on range. Last year that aircraft managed a successful air-to-air refueling from a manned aircraft.
In that achievement, the aircraft now had to compensate for THREE bodies moving in three dimensions – the X-47 itself; the manned refueling aircraft; and the refueling basket at the end of a flexible hose, which the X-47’s refueling probe had to engage.
Adding that third body to the equation complicated the challenge by orders of magnitude.
For this accomplishment to fulfill a practical utility for the Navy, try to imagine what has to happen.
The cognitive systems controlling it need to be able to prioritize the importance of the particular mission – which is variable – against the risks to the human beings aboard the manned refueling aircraft.
Those systems need to be able to factor in:
- The roughness of the air;
- The time that the refueling aircraft can spend on station trying to complete the procedure;
- The threat to the human flight crew from whatever dangers are present;
And a whole host of other factors, which are changing from moment to moment; and with each change, affect the priorities of the other factors in a non-stop domino effect.
The challenge was immense.
So how do these cognitive systems work and why do they pose such an enormous challenge?
First, let me address a misconception – an understandable one, but a misconception nonetheless.
Many non-engineers often presume that technology is constantly progressing with analytical continuity, where future results simply build on the results of the past.
We all understand Moore’s Law – that processing speeds have doubled every eighteen months or so.
And I think this is one reason why so many take technology’s progress for granted. We tend to presume that any computing-based problem can be solved if we are just willing to wait for Moore’s Law to catch up to our ambitions.
The development of manned aircraft in recent years would seem to support that.
In 1974, the first model of the F-16 included 135 thousand lines of code. In 2006, the first model of the F-35 included about 7 million lines of code. Today, the latest model of the F-35 includes about 24 million lines of code.
Originally – about 30 years ago – the B-2 bomber could hit eight targets independently per mission. Today, it can hit 80.
But here’s the difference: Those aircraft are manned.
Most of the cognitive needs are supplied by the pilot – highly trained and practiced. Without the pilot, processing speeds – though still necessary – are suddenly inadequate.
Without the pilot all those moment-to-moment judgment calls need to be supplied by software – specifically by algorithms.
An algorithm is a very simple thing in concept. It would appear to be nothing more than a set of rules designed to allow a computer to solve a specified problem.
But today’s algorithms need to learn, and apply that learning to solving unexpected problems. Real operational environments are stochastic by nature – and the algorithms used must deal with that variability, unpredictability and inherent randomness of the real world.
So much of the advancement underway for autonomous systems isn’t about the linear progression of hardware capabilities. We are now deep into the world of computer science and we are seeing nonlinear acceleration of capability with the advancements in this field.
Tremendous progress has been made. In fact, the X-47B has demonstrated an ability to operate in conditions that would prohibit a human pilot from taking to the air.
And one telling measure of its reliability is this: On landing, it touches down on precisely the same point of the flight deck; so reliably that the Navy has asked us to program some randomness into its landing performance to keep that part of the flight deck from prematurely wearing out.
Of course, tremendous progress is still needed.
In the case of the X-47B, the next challenges to overcome will be to prove it can accomplish its missions despite jamming attempts or a loss of GPS.
It will need to prove it can prevail over hostile countermeasures and cyber-attacks, in confused environments that include both friendly and hostile elements.
And above all, it will need to hold the confidence of our policy-makers – and the citizens who empower them – that they will operate reliably and in a manner consistent with our values.
Quite a high bar. And, as with any serious R&D initiative, there will be setbacks and unforeseen discoveries to address. But I am confident that these challenges will be solved.
And their solutions will prove to be a boon to aerospace and to the human condition in general.
We can say this with confidence because we already see the contributions such systems are making.
The Global Hawk autonomous aircraft has performed years of service gathering near real-time intelligence, surveillance and reconnaissance information.
But it has also performed humanitarian missions over areas devastated by natural disasters such as Haiti, the Philippines, and Japan.
And it has generated tremendous scientific data on hurricanes for NASA.
It’s safe to say that the defense industry has long been a clear net exporter of capabilities to the commercial world. Think of GPS, computer networking, prosthetics, satellites and communication technologies.
Many of those spillovers into the commercial world catalyzed the expansion – and even the creation – of entire industries generating astonishing human benefit.
And fully-realized autonomous aircraft will be no different. It’s easy to foresee tremendous benefits for commercial aviation. Benefits in the way of capability, operating costs, efficiency, and above all, safety.
But whatever one thinks of this technology – however skeptical or suspicious one might still be – it represents a genie that is well out of the bottle.
Nations around the world are working to overcome these same challenges for the same purposes, including China and Russia.
And though the United States is still the leader in overall research and development spending, that is changing fast.
As an example, measured in Purchasing Power Parity (PPP), Battelle forecasts that China’s total R&D funding will surpass the U.S. by 2022.
Furthermore, the quality of R&D in China is impressive. Chinese researchers are increasingly publishing in top rated journals and may soon dominate research in a number of fields.
China’s R&D budget has grown faster than its GDP growth. In the last five years, its R&D spending grew about 18 percent annually, double its historic GDP growth rate.
In contrast, over that same period, total U.S. R&D spending grew annually by an average of only 1 percent while defense R&D declined by an average of 7 percent each year.
We are clearly going in the wrong direction.
Not only are we limiting ourselves with lower R&D investments. The progress of this technology is outpacing the policies needed to manage it. It’s a situation that acts like a brake on its progress and even serves to place our nation’s technological lead in jeopardy.
As the President’s National Security Strategy, published last year, states: Innovation, "empowers American leadership with a competitive edge that secures our military advantage."
But innovation doesn’t simply happen. We know that our efforts at R&D must be carefully managed and cultivated. Particularly under today’s fiscal restraints.
In this, the age of innovation, technology and human intellectual capital, one of the most important choices we can make is to do those things that support and advance R&D and innovation.
For those of us in the technology industry, there can be no greater satisfaction than being associated with the design, construction, and refinement of a complex piece of technology that eventually becomes indispensable in the affairs of people.
Not every technological innovation can make that claim.
But I believe that autonomous unmanned aircraft systems, as a class of technology, can and will.
Ultimately, they will not be confined to military uses. Already, they are shaking off their identifiers as novelties and curiosities. Every year they grow more reliable, capable, versatile, and affordable.
Autonomous aircraft continue to prove their value and their potential and I am certain of a day – not too far off – when they will be safely integrated into the fabric of our security and our economy.
I am excited about the work we are doing to bring this to reality, and look forward to the many positive impacts this technology will bring.
# # #