Behind the Screens: Developing Flight Controls for Bell’s Nexus Air Taxi

Designing flight controls for a new helicopter usually begins with the traditional layout of a collective lever to the left, cyclic between the pilot’s legs and two rudder pedals on the floor.

More options are available for airplane controls since the advent of fly-by-wire systems, but the basic layout has remained the same for decades. There is a control mechanism — either a side-stick, a center-stick or a yoke — to signal the appropriate actuators as well as a throttle quadrant and rudder pedals.

With the digitization of systems and increased automation, there’s no technical reason that flight controls need to look like that other than pilots expect it. A helicopter pilot is used to flying a helicopter with rudder pedals and a cyclic, so manufacturers keep making them, perpetuating the cycle of cyclics.

The urban air mobility (UAM) market represents an opportunity to revamp the idea of a vertical-flight cockpit. A whole new industry is developing and companies don’t yet know what the aircraft will look like, who will fly them or how.

Bell doesn’t think the current crop of helicopter pilots is sufficient to staff the UAM market, because there already is a shortage in filling current pilot positions for a narrower set of missions than UAM purveyors envision. As a result, there is no need to tailor flight controls to the expectations of existing pilots.

Bell started a round of hands-on research with three simulators used to collect data on how future pilots will control VTOL aircraft. One has traditional helicopter controls based on Bell’s Jet Ranger: cyclic, collective, rudders. One has two side-sticks and no rudder pedals. The third one is pared down to just one side stick on the operator’s right side.

“What you don’t see is, behind the scenes, there is a very powerful database system collecting hundreds of pieces of information 20 times per second,” said Bell experimental test pilot Jim Gibson.

Bell takes the simulators to trade shows, such as the Consumer Electronics Show and Uber’s Elevate Summit, as well as internal Bell events and outreach programs it does with high schools and colleges. It collects demographic data on participants and then tracks their performance as they fly a simulated air taxi through a timed, zig-zagging course down the Las Vegas strip.

Participants are then asked to answer a series of yes-or-no questions designed to quantify the workload they experienced.

The post-flight questionnaire is based on the Bedford Workload Scale that test pilots use, Gibson said.

“It’s a series of questions you ask yourself to give a quantitative score to what is a qualitative task,” he said. “What I did was take that scale and translate it into a series of questions that a person who isn’t a trained test pilot would be able to understand and go through with a yes-no.”

Bell is not ready to release results, but Gibson said there have been surprises.

“A lot of people’s preconceived notions about who might be successful and who might not be in different configurations has not been correct,” he said of the roughly 5,000 participants Bell has tracked so far. “An older woman came who had never even flown in an aircraft before. She had probably one of the top three scores we had ever had. A confident airline pilot was probably our worst performer.”

When choosing how to design the simulators, Bell considered a lot of options. Setting one up like a helicopter as a baseline was an easy decision, but the other two required research, Gibson said. He had the idea of scrapping rudder pedals because they are a common bugaboo for novice pilots across fixed-wing, tiltrotor and helicopters, but other than that, it was about learning from available studies.

“Before I came to Bell, I was in charge of V-22 flight training for the military,” he said. “I look at it as, how can I train [someone on an aircraft] quickly with the same level of safety.”

After several months of reviewing journal articles, one trend that emerged was adopting video game controllers. Bell has stayed away from direct adoption of, for example, a Sony Playstation 4 controller, but did incorporate lessons from those controllers.

“You can give really fine inputs when you’re sitting on a couch. When you’re in the air, it’s much harder to make fine control inputs, because the ergonomics of your fine motor controls switch,” Gibson said. “When you look at Gameboy controllers or drone controllers, or even the average backhoe, what are the three things you see on them? They’re fairly simplistic. They have two joysticks or thumbsticks … and a very limited number of buttons,” Gibson said. “So, the designs we’ve included, we did take that concept of two inputs to heart.”

Bell’s two-stick format has a right stick for controlling pitch and a left stick for controlling throttle and yaw. It likely will feel familiar to anyone who has frequented an arcade. On the single-stick setup, which features one control column to the operator’s right, the throttle is moved to a thumb dial and the left hand is free for other tasks.

Through the Elevate Summit’s June show, Bell has been in what it calls “phase one” of the flight control testing program. Soon, it will kick off phase two, according to Carey Cannon, the chief engineer for part of Bell Innovation, which works on UAM and electric vertical-takeoff-and-landing (eVTOL) technology.

Phase two will be broad, likely lasting 18 months to two years or more. There Bell will introduce flight control laws and basic artificial intelligence functions, Gibson said. The main point of phase two is identifying how much of a pilot’s workload can be automated during basic flight modes. Ideally, the pilot is unable tell the system is interceding. The automated flight control computer operates in the background, making his or her job easier.

“Let’s say I find it is difficult to hover the aircraft and maintain both altitude and direction,” Gibson said. “If there are things we can do to help a person put a vehicle in a hover and then constantly jockey the power in a hover once they get there — it is easy, but is it easy for someone new? The data we collect will help allow us to develop control loops behind the scenes.”

“Behind the scenes” is key there, because Gibson isn’t talking about additional options. These aren’t modes a pilot can toggle on. They are always-on systems running without any input from the pilot.

As part of phase two, Bell will up the simulators’ data-gathering power from 20 data points per second to 50, also capturing individual control inputs to help understand where people are having difficulty and whether the added automation is smoothing things out. Rather than a static data collection period, phase two will be iterative, so Bell will continually monitor and make changes to see what works best.

Phase three doesn’t have a specific date but will focus on optimizing the operators’ interaction with the cockpit, ensuring that the pilot has access to the necessary information and displays and nothing more.

“My personal thought is sometimes all this information can be a detriment,” Gibson said. “You can see with other [manufacturers,] as systems start to degrade because of malfunctions, we need to take information away from [the] pilot to get them to concentrate on what’s there, because that’s what they need to safely operate the vehicle.”

Gibson believes that a capable autopilot can monitor data such as system temperature and engine oil and just let the pilot know if a problem needs attention.

“Modern autopilots are excellent at deterministic behavior,” he said. “You know that when the temperature reaches a certain point, you need to do X.”

If pilots don’t need to have constant access to that information, that’s a good way to simplify the vehicle operations. Cannon said operators could become more like monitors, overseeing trips while data is collected in early operational phases to help certify unmanned operations. Automated road vehicle operator Waymo (formerly part of Google) does the same thing, and Starsky Robotics is applying this model to the trucking industry.

With a roadmap extending years into the future, this project will not be completed by 2025 and the planned launch of Bell’s Nexus eVTOL., though the aircraft is one of the primary drivers for the project, according to vice president of innovation Scott Drennan. Early returns from the program will be integrated into the Nexus’ controls, which are being developed jointly with Thales.

Garmin is developing vehicle management computers for Nexus. The two partner companies will need the information Bell sends them, but don’t need anything immediately to work on their pieces for the Nexus, Drennan said.

“Garmin expects to adapt not only its contribution to the flight control system, but also the human and aircraft user interface,” Vice President of Aviation Programs Systems and Business Development Didier Papadopoulos said in an email. “Because the Bell Nexus utilizes an unconventional flight control system, this is something Garmin is diligently investing in and working through with our partners.”

While the Nexus is the most notable use for Bell’s future flight controls initiative, it isn’t the only one. Part of the reason the research program will extend beyond the air taxi’s launch is the company plans to use what it learns to improve the intuitiveness of other aircraft flight controls.

Uber’s business case hinges on scaling up to thousands of air taxis operating in a city. With an expected need of about 2.5 pilots per vehicle, Uber anticipates requiring around 5,000 pilots in just its first five years of operation, according to Engineering Director of Aviation Mark Moore.

The ultimate goal of Uber and partners such as Bell is unmanned operation to cut the need for pilots, dropping a big expense and freeing up an extra seat for paying customers on each ride.

Most observers of the industry — including acting FAA administrator Dan Elwell — don’t expect that to be feasible right off the bat. With early entrants wanting to start service by 2023 to 2025, pilots will be a necessary evil in the early going.

“We, I, the FAA, understand your desire to sprint out of the starting gate,” Elwell said at Uber’s Elevate Summit. “But you have to understand our safety [regulation process.] When we’re ready, we’ll move to urban areas with semi-autonomous operations.”

But pulling thousands of pilots out of the already workforce-strapped commercial industry, especially if they have to be helicopter-certified, isn’t realistic. That’s why Bell wants to create a new type of pilot, or operator, that is easier to certify.

When it’s time for unmanned flight, Cannon said some of those operators can transition into new roles as ground operators, an air traffic controller-like role, or in data analytics. The immediate concern is figuring out who those pilots will be and where they will come from.

“It’s not that we’re going to train for gamers or drone operators, but what it allows us to do is accommodate gamers, accommodate drone operators, accommodate people who have never seen an aircraft before,” Gibson said.