Pasadena PD Air Operations, UK SAR photo feature, 2019 SAR Training Directory, part 3 of lessons learned in risk management, and more!
As engineers and researchers from across the industry continue their ongoing quest to refine and enhance the next generation of rotary-wing aircraft, one key theme has emerged from the changes planned for the cockpits of tomorrow: a reduction in pilot workload. Primarily driven by advancements in interfaces, arriving between 2025 and 2030, helicopter crews will be able to allocate more time than ever before to mission management. The same period will likely also see a growth in the number of types using fly-by-wire controls, but while artificial intelligence (AI) is being intensively studied by many companies, it has yet to work its way into a certifiable decision aid.
“The strongest design driver is user interface,” Jim Gibson, an experimental test pilot with Bell, told Vertical. When the complexity of a flight increases, for example, a pilot may become overwhelmed and lose track — and this is when controlled flight into terrain happens. To help avoid this, “let’s make it easier for the pilot to operate the aircraft,” said Gibson.
With less time spent on the physical aspects of controlling the aircraft, the pilot is able to spend more time focusing on the operation at hand, he added. “In a medical evacuation, that would be coordinating with the nurse in the back, communications with air traffic control and the hospital, as well as coping with challenges such as night or weather,” Gibson said.
Jon McMillen, in charge of business development for Sikorsky Mission Systems, is on the same page. He sees the pilot as a mission commander, monitoring the flight and making inputs when relevant. This is what drove the development of the in-service automated rig approach feature on the Sikorsky S-92, he said.
In the future, automation might see an aircraft take off and fly to a waypoint, avoiding obstacles, before hovering to pick up a patient and a physician. The pilot would therefore have mental resources available to better plan ahead. “You want the pilot fully engaged when he gets there,” said McMillen. In an operation such as firefighting, crews often have to deal with unexpected situations, and automation of the more procedural aspects of the flight could allow them to increase their focus on the more challenging aspects.
Sikorsky has conducted extensive research in that direction with the Sikorsky Autonomy Research Aircraft (SARA) demonstrator, an S-76 fitted with special equipment to give it various degrees of automation (see p.56 for our flight report). “We demonstrated a helicopter can be flown with a tablet — without conventional gauges, horizon line etc. — using just a flight profile,” said McMillen. “There is a paradigm shift on how raw data can be replaced with a mission or flight intent.”
Full autonomous flight is far off, especially because of certification challenges. But the interface Sikorsky has created for SARA may find applications in conventional rotorcraft.
“The information is becoming more consolidated, easier to digest, with less raw data,” said McMillen. An automated checklist may include tests running on their own, using oil pressure sensors for instance. “You can have the system run it through the startup process, pointing to a system you need to look at and interrogate,” said McMillen.
New ways of displaying information may help. The pilot in a conventional cockpit has to aggregate numbers from the variometer, altimeter, speedometer and fuel gauge, as well as engine parameters. “You can do that for him, a bubble — varying in size — around the helicopter’s symbol [on a display] can represent the available power and its color may give a clue on the fuel level,” said Christophe Bey, executive director of Akiani, a French company specializing in human factors and user experience.
Reducing the pilot’s workload saves cognitive resources, and greater mental availability leaves room for preparing action plans, Bey told Vertical. “Sometimes a pilot receives so much information that he tries to understand things that are not worthwhile — this is a human bias — and forgets anticipation tasks.”
In terms of hardware, transparent avionics screens may give better visibility when close to the ground. Opacifying panels may recreate the instruments, Bey suggests. Some of them would be switched off for takeoff and landing.
The use of autopilots may become widespread in the future. The use of such systems has already been democratized to an extent — the Airbus H135 light twin has been offered with a four-axis autopilot for a few years, while the Genesys HeliSAS is bringing complete two-axis autopilot functionality to a growing number of light aircraft types. Avionics giant Garmin is looking at bringing further automation to lighter helicopters, and hopes to cut its cost. Referring to stability augmentation systems and autopilots for visual flight rules (VFR) helicopters, Bill Stone, Garmin’s senior manager, aviation business development, points out the amortization challenge. “You can design a low-cost solution but you have to amortize it on a small market,” he said. “We try to find scalable solutions to use on multiple platforms; we manufacture in-house to tightly control costs.”v
Another means to reduce pilot workload may be through the use of voice command. “Technology has really come along in the last few years,” said Stone. “For routine tasks, the pilot can simply talk to the avionics.” In the terminal area of an airport, for example, a vocal order for a wind check is easier than manipulating the avionics, he suggested.
Voice command belongs to a trend for increased resilience of the system/human interface. The more intuitive the interaction, the smaller the risk of error, said Akiani’s Bey. “Tapping on a touchscreen is better than typing coordinates,” he said.
Fly-by-wire controls may help reduce workload, too. Over the last two decades, however, they have been a tantalizing goal for design engineers at helicopter OEMs, with the cost making a business case harder to prove than with their fixed-wing counterparts. Airbus Helicopters, for instance, pursued the technology for civil applications, considered it for the then-named X4 program, but eventually dropped it for the resulting H160.
Military rotorcraft such as the NH90 and the CH-53K do have fly-by-wire controls. But the first civil certified helicopter is set to be the Bell 525 Relentless super medium twin. “Fly-by-wire costs have come down over the last 20 years in the military,” said Bell’s Gibson. And he is certain that costs will continue to fall, allowing such a system to arrive in lower-cost helicopters. “The better we become at it, the cheaper it gets,” he said. “At some point it will become compatible with a light twin.”
Sikorsky’s McMillen agrees. “Fly-by-wire technology has been maturing and cost is coming down,” he said. “We will see it on larger [commercial] aircraft sooner rather than later.”
Dan Toy, principal business development manager for avionics at Collins Aerospace, said that making fly-by-wire affordable is a challenge, largely because controls have to be triple redundant, and this extends to every single element. “You have to make the system work under any type of failure, so it increases complexity,” he said.
All manufacturers agree fly-by-wire certification is very costly. But such controls have “tremendous benefits,” said Toy. The capabilities of an autopilot are much greater if coupled with fly-by-wire controls, added Akiani’s Bey.
On the Bell 525, pushing the cyclic stick forwards or backwards directly controls speed. The pilot does not have to mentally reckon what the pitch angle should be, relative to the desired speed and the aircraft’s weight. For takeoff and approach, the system enables repeatable maneuvers, whatever the weight.
In case of a double engine failure (which statistically might happen once in an aircraft’s life), the pilot usually has to react within a handful of seconds. This is the responsiveness needed to keep the rotor turning in order to perform an autorotation. Fly-by-wire controls will give the pilot a head start, automatically applying the first of the required control inputs.
The use of fly-by-wire also offers new possibilities in the design of controls. Engineers may take a hard look at how the pilot interacts with the aircraft. Bell is thus testing a variety of physical configurations for flight controls in its future vertical takeoff and landing (VTOL) aircraft, including the Nexus. About 800 people — including both experienced pilots and non-flying consumers — have trialed Bell’s three different configurations through virtual reality simulators, according to Gibson. “We are in survey mode,” he said. Bell will start working on flight control laws next year.
Increasing situational awareness
Enhanced and combined vision systems are perhaps closer to maturity. Enhanced vision relies on optical sensors (such as infrared cameras) to enable a crew to see through fog or at night. Flight data, such as speed and an artificial horizon, is superimposed on the picture. Combined vision adds synthetic terrain.
While such systems have proved to improve situational awareness in the fixed-wing industry and in military rotorcraft, they have yet to find their way into civil helicopters. Civil fixed-wing aircraft mostly use head-up displays. Due to their lower speed, wider field of vision and varied operations, helicopters would likely be better suited to head-worn devices. These are fairly common in the military world, Toy noted. But making those devices affordable is a challenge, which Collins Aerospace is working on.
Thales is developing a monocular system. This will allow crewmembers to share their view of items of interest, such as an accident site during an air medical mission. One crewmember could be looking at the site on the ground, and their viewpoint could be shared on another crewmember’s display. On such a system, the next waypoint may also be highlighted, Bey suggested.
AI may sound like a buzzword for cockpit improvement, but the kind of AI technology that has made spectacular progress in the consumer world is, to date, unsuited to cockpit electronics safety standards.
Based on neural networks, a Siri-like AI system learns from a large amount of data. It works in an empirical way, but it is capable of making an error. Moreover, the system cannot explain its choices, thus undermining the confidence a human operator may want to place into it.
In France, experts at Thales and aerospace research center Onera disagree on how long it will take to design a certifiable AI system — their respective predictions being a minimum three and 10 years. Collins Aerospace’s Toy references “the late 2020s.”
Ideally, AI could become “a competent assistant,” as Bell’s Gibson put it. “Pilots do not see the autopilot as a copilot for good reasons — it is deterministic and may lack a response to an unexpected situation,” he said. But the probabilistic (as opposed to deterministic) nature of AI makes its certification virtually impossible with the current approach. Certification authorities will have to change the way they look at a new system, said Sikorsky’s McMillen. An airframer will have to prove to the certification authority that the machine can be trusted. An application may be, in the case of a worrying alert in the cockpit, helping the pilot to find an alternative airfield.
Trust is already an issue with current automation. “When the situation gets worse, some pilots no longer have confidence in the system, such as an unexpectedly disconnecting autopilot,” noted Akiani’s Bey. The pilot distrusts the system for the rest of the mission, believing it has let him or her down. They should put the disconnection in context, which would make them realize the helicopter’s attitude, for example, was outside the autopilot’s operating envelope, said Bey.
An algorithmic (i.e. deterministic) form of AI exists, but “is 10 years behind” probabilistic AI, according to Bey. Some advanced automated systems incorporate algorithmic AI. “Our auto rig approach feature is a form of AI: it knows your goal, flight regime and brings you there safely,” said Sikorsky’s McMillen.
In addition to adapting the cockpit to human beings, what about better selecting the latter? Progress in neuroscience has enabled the creation of new tests. They can help select those persons whose behavior degrades less under stress. Or, the cognitive state of the pilot may be checked before flight, said Bey. A major challenge, however, would be how to manage the result if a pilot is found unable to complete the flight — has the operator got a replacement pilot ready to go?
While many fear that increasing automation may eventually make pilots obsolete, they shouldn’t worry too much about the future of their profession, said Bey. It would be extremely expensive to replace them with the required redundancies, he said. Keeping humans in the cockpit will continue to make sense for the foreseeable future, with automation simply serving to help increase the safety of their work.