In conjunction with the Pentagon’s Defense Advanced Research Projects Agency (DARPA), Sikorsky Aircraft (stand 8209) recently conducted the first autonomous, uncrewed flight of the S-70 Black Hawk. Last month’s landmark mission lasted 30 minutes at the US Army’s Fort Campbell base in Kentucky.
The unmanned helicopter was part of DARPA’s Alias (for Aircrew Labor In-Cockpit Automation System) program. Sikorsky’s Matrix hardware and software autonomy technologies are at the heart of Alias, and Igor Cherepinsky, Director of Sikorsky Innovations, is at the heart of Matrix.
In addition to military applications such as Alias, Cherepinsky sees Matrix as a game-changer for high-risk civilian missions, including nighttime aerial firefighting and search and rescue (SAR) over the air. ‘water. Lockheed Martin subsidiary Sikorsky is currently working with the FAA to certify the system, which uses a plethora of onboard sensors, lidar and cameras coupled with proprietary hardware and software.
The system comes with a kit that converts virtually any aircraft into varying degrees of additional electric flight control that Cherepinsky likens to “lane assist” on cars. “Depending on what kind of steering it has, some cars work well with lane assist and some don’t,” he said.
The key to Matrix’s “smooth operation” is Sikorsky’s proprietary algorithms, which differentiate it from existing drone-based guidance systems that Cherepinsky calls “sophisticated waypoint-following autopilots.” Matrix can be flown in both line-of-sight and beyond line-of-sight, and an aircraft equipped with it can have its mission changed from an operator “in the cockpit, aft, to Oklahoma, or on the moon,” he said.
With Matrix, an operator gives the system mission objectives and constraints, “and the machine produces a plan that it will follow. If you can’t talk to him, he’ll still produce a plan and he’ll handle contingencies if anything deviates from the norm,” Cherepinsky said. “It’s not some kind of ultimate intelligence, but it’s pretty robust.
For example, while Matrix is constantly running through the aircraft checklist, if something falls outside the minimum in-flight equipment list and is mission critical, the system can land the aircraft quickly. “We demonstrated this back-to-basics functionality and will be doing more demonstrations of this soon,” Cherepinsky added.
Despite all these features, he said Matrix does not rely on artificial intelligence in terms of machine learning when critical flight decisions need to be made “where failure would mean something terrible would happen”. Nor would any military applications be used to make weapon decisions.
However, he said it could be used to help target “for safe aircraft operation”. For example, “If the aircraft is in a degraded visual environment, Matrix will ensure that the aircraft can continue to fly safely, but completely independent of weapon systems.”
Matrix is designed to be modular from both a hardware and software perspective for ease of maintenance and updating, and the architecture is scalable up to eVTOLs or “even smaller aircraft than that”, a said Cherepinsky. But he added that the system cannot be updated wirelessly.
“Technologically it could, but because of cybersecurity and other concerns it probably shouldn’t be, so it’s not designed to do that,” he said. But Cherepinsky added that third parties will be able to write apps and fly the plane.
Sikorsky is also exploring a wider window of human-machine interfaces for the system, including the possible development of wraparound one-piece touchscreen cockpit displays that are simple and intuitive. Cherepinsky hinted that Sikorsky is developing such a display for its Matrix S-76B Sara (Sikorsky Autonomy Research Aircraft) platform and has also performed demonstrations with Matrix using voice commands for secondary functions for tasks, including the changing radio frequencies and performing checklists.
But it’s Matrix’s ability to speed up SAR missions over water that is perhaps its most impactful utility, what Cherepinsky calls “super search.” Rather than setting up an autopilot to fly grid patterns, Matrix can be instructed to search for a boat, person, or something even smaller.
“Matrix knows what sensors you have on the plane and builds a map model of them. It understands the environment and actively creates a research pattern which to a human may seem random, but which in flight tests and simulations [can find the target] about 30 percent faster compared to [flying] a typical search pattern,” he said.
Notably, Matrix can calculate an object’s drift through water over time, locate and calculate the approximate search area, and then use lidar and cameras to locate the object. “If someone is in the water, every minute counts,” Cherepinsky said.