Military, Products, Regulatory

Sikorsky Matrix: Advancing Autonomy to Higher Levels

By By Dale Smith | May 6, 2014
Send Feedback

Sikorksy Matrix Technology at AUVSI Unmanned Systems in 2013. Photo by Woodrow Bellamy III

It seems like autonomous helicopters or unmanned aerial vehicles (UAVs) are all the rage these days. From individual UAVs being turned into package deliverers to a swarm of small quad-copters playing the James Bond theme, it seems like there is no limit to what we’re going to ask these small rotorcraft to do.

And while, from a technological standpoint, many of these accomplishments are pretty amazing, other than a public-relations grab or pure entertainment, few are of any measurable benefit.


Thankfully though, company founder Igor Sikorsky’s dream of giving vertical takeoff and landing (VTOL) aircraft – even those operated without pilots on board – the capabilities to serve mankind’s greater good is alive an well in the form of the Matrix Technology.

Matrix is Sikorsky’s major research program to develop, test and field systems and software that will significantly improve the capabilities, reliabilities, and safety of flight for autonomous, optionally piloted and piloted VTOL aircraft.

“What we are basically trying to do is to automate a lot of things in current VTOL machines that the human (pilot) now has to do,” explained Igor Cherepinsky, chief engineer on Sikorsky’s Autonomous Program. “You can think of it like the old days when you got on an elevator and there was an operator there to run it. It was too complicated for an untrained person to do safely.”

“As automation advanced, you did not need that operator any longer. You just push a button,” he continued. “We are working to bring the same advancement to the pilot/aircraft interface. We’re saying it doesn’t really matter where the human being is – but there needs to be human interface at some level – but that human can either be in the aircraft or at a remote location. It doesn’t really matter.”

Cherepinsky said that with Matrix, instead of the pilot having to worry about sticks, rudders, cyclics, collectives and the like, he is able to concentrate on “managing the mission.”

This is ultimately what Matrix is all about, he explained. “But like any new technology it has an upward spiral development path. We need to start by making the machines autonomous and more reliable at lower levels (capabilities) and build up from there.”

As Sikorsky’s vice president of research and engineering, Mark Miller stated in a recent presentation: “The game-changing Matrix Technology we are developing and testing will provide orders-of-magnitude improvements in system intelligence and contingency management to ensure high levels of reliability, and ultimately, make unmanned missions by helicopters and other VTOL aircraft of all sizes highly affordable,” he said.

Mike Miller with Sikorsky Innovations.
Photo courtesy of Sikorsky

Both Cherepinsky and Miller stressed that improving the aircraft’s reliability and, more importantly, safety are the foundation of the program’s efforts. Of course if any of the autonomous aircraft have read Isaac Asimov’s Three Laws of Robotics: much of this would not be a problem. But we know that as smart as they are, autonomous vehicles can’t read. But, I digress…

Cherepinsky explained that the current industry loss rate for unmanned aircraft is right around one-per-1,000-hours of operations. That’s way too high to allow any meaningful UAV applications.

“They’re maturing but most are single- or dual-strand systems. They don’t have enough redundancy to really sustain high-tempo lossless flight,” he said. “Our goal is to improve that loss rate to one-in-100,000 flight hours. The combination of our robust architecture, multi-level contingency management and advanced system intelligence algorithms will enable that transformation.”

He continued: “We have to bring autonomy into the vehicle in a very reliable fashion and to enable the pilot to become a mission operator. Pilots are never going to disappear, but their role is going to change. There are lots of missions where humans don’t really belong on the aircraft. But, there are also lots of missions where you really do want human eyes right there in the middle of the action.”


Developing the First AI Autopilot

So by now you’re thinking, okay the “pilot” can be located hundreds, if not thousands of miles away from the Matrix-equipped UAV, that’s nothing special. Yes, they both can operate remotely or even autonomously, but Cherepinsky stressed that that’s pretty much where the similarity ends.

“Today’s UAVs borrow a lot of technologies from the autopilot systems used by both VTOL and fixed-wing applications. They are all waypoint-oriented in the way they operate,” he explained. “You enter a flight plan and it goes off and flies the mission. That’s all fine as long as the mission is that simple and nothing goes wrong.”

Matrix is the first system “that really allows the operator to modify the mission and mission goals,” Cherepinsky stated. “For example, the operator can just say I need to get from Point A to Point B – what happens between those two points becomes totally up to the machine.”

If weather or mechanical problems pop up during the mission, the Matrix-enabled UAV is free to modify its flight plan to take these issues into account when continuing the mission or opting to land.

“The machine itself will be able to handle all major contingencies. So if there is a single-item failure, it will be handled on board. If there is something the machine cannot cope with it will notify the operators,” he said. “Instead of telling you that ‘engine one is out’ it may just say that its range or payload capacity has gone down or give you other immediate affects on the mission. It’s up to the pilot/mission manager to take it from there.”

Even if your UAV is remotely-piloted, if there’s a mechanical issue and it needs to land – now – it’s nearly impossible for a “pilot” who is time-zones away to know if there’s a suitable landing site within range. From Sikorsky’s point-of-view, it would be easier, and safer to give that autonomous, task aware decision-making capability to the aircraft itself.

One of the key capabilities of the Matrix Technology is the ability for the system to autonomously find a suitable site and land the aircraft.

“The basics of any truly autonomous helicopter is it needs the capability to understand its location, obstacles, and terrain and can actively plot a path that enables it to land in the sorts of places helicopters need to land in both normal and emergency situations,” Cherepinsky said. “Pilots do it all the time. They go out to a site, locate the obstacles and plot a path around them to land safely. This is one of the first capabilities we need to give to our machine.”

“At the end of a mission or during an emergency the ‘pilot’ can either make the final decision or just let the machine proceed with no intervention,” he said. “That’s really the first spiral of true artificial intelligence and autonomy that we are starting to put into our products.”

How does it work? Well, Cherepinsky said it’s still top secret and if he told us he would have to … well, you know the drill.

Anyway, while a totally self-operating helicopter may seem the stuff of science fiction, and depending on the FAA, it may well be for a long time yet, Cherepinsky said that autonomous site selection and landing capabilities are an excellent addition to the cockpits of crewed helicopters.

“One application could well be a ‘Please Land Me’ mode,” Cherepinsky explained. “Should a pilot become incapacitated or disoriented, he could engage the system, which will use its onboard perception system to examine the immediate areas and locate a suitable landing spot. Once there, the pilot can choose to handle the landing or the system can do it.”

“It’s just like the autopilot flies a set flightpath today, but this is much more elaborate and intelligent,” he said. “We see this as a major step in eliminating CFIT (controlled flight into terrain) accidents. We don’t want to see another machine that is perfectly capable of flying being flown into the ground because the pilot is disorientated or cannot see.”


SARA Platform

Sikorsky autonomous research aircraft flying over New York. Photo courtesy of Sikorsky

To help put all of the Matrix Technology theories into practice, Sikorsky’s engineers chose to buck the small aircraft trend and go with a full-size S-76 as its test and demonstration platform.

“We had done some work with small remote controlled platforms, but those tend to be computationally and payload constrained and are not really representative of a real aircraft,” Cherepinsky explained. “Having the large aircraft like the S-76 has been very good for us. Using an aircraft that looks and performs like our objective vehicles enables us to tackle problems that are real and not waste time dealing with issues that only show up in smaller aircraft.”

With its new fly-by-wire conversion and Matrix Technology installed, the SARA S-76 is enabling rapid flight-testing of all of Matrix’s hardware and software, including the multi-spectral sensors.

To date, Cherepinsky said Sikorsky has completed Phase 1, while Phase 2 of the Matrix’s spiral development process is progressing very well.

“Phase 1 demonstrated that we can provide full autonomous operations and very accurately control the vehicle to put it where it needs to be in both position and time,” he said. “We can easily hold hover that stays within a one-foot sphere in winds gusting from 12- to 28-knots.”

Cherepinsky said that Phase 2 includes the first demonstration of Matrix’s AI capabilities. “We’ve installed a multi-level sensor system on the aircraft and that’s what we’re using to locate and avoid obstacles around landing zones,” he said. “We had some big successes with it. We recently flew over an area and identified the landing zones. We didn’t land yet. We were just verifying the algorithms. We are planning a series of other demos over this year with increasing complexity in these areas.”

Cherepinsky said while a full system roll out is probably years away, he does see situations in the near-term where particular applications of the Matrix Technology “suite,” including the aforementioned “Please Land Me” mode, can be put to use by commercial and military operators in the very near future.

“Of course in the near-term the major technological transfer will be in more sophisticated autopilots for helicopters and fixed-wing aircraft,” he said. “Eliminating CFIT is a big priority. Further, there are applications for more advanced optionally-piloted and autonomous aircraft. That may take a while – not because of technology, but more from a political and certification perspective.

“We’re having ongoing conversations with the FAA on this technology and how we will roll it into the system,” Cherepinsky said. “The FAA has been doing an excellent job trying to find ways to get UAVs integrated into the NAS (National Airspace System). It’s going to take a bit to get it all done.”

Matrix Technology “addresses the unique needs of vertical flight systems,” Miller said. “By our efforts we seek to expand the types of missions that can be flown, improve the efficiency of existing missions and continue to build on the safety and reliability that has been a Sikorsky hallmark.”

Commercial | TechnologyMike Miller with Sikorsky Innovations. Photo courtesy of Sikorsky Sikorksy Matrix Technology at AUVSI Unmanned Systems in 2013. Photo by Woodrow Bellamy IIISikorsky autonomous research aircraft flying over New York. Photo courtesy of SikorskySikorsky MatrixCommercial | TechnologySikorsky Matrix

Related: Unmanned News

Receive the latest rotorcraft news right to your inbox