Here’s How the Autonomous Air Taxi of Tomorrow Won’t Let You Die

The Drive’s tech writer and aviation expert takes us into the future for a ride in a self-flying electric aircraft. It might be a bumpy ride.

byEric Adams|
Self-Driving Tech photo
Share

0

NEW YORK CITY—2052

The wind and snow were pounding against the skyscraper window beside his desk when Steven—dealing with a chaotic array of challenges at his Midtown office—felt a slight thump in his ear preceding a soft voice only he could hear, reminding him that he needed to be at JFK Airport in 30 minutes to make his flight to San Francisco.

“What terminal?” he whispered. “Park & 32nd,” the female voice responded.

Just a decade prior, anxiety about traffic and the other hassles of NYC-area transportation would have kept his airport dash front-of-mind all morning. But times had changed. Hundreds of air taxi ventures had gone into financial tailspins over the past three decades, along with quite a few actual tailspins that had resulted in some extremely well-documented fatal accidents while testing the new aircraft. (Blame the massive technological hurdles relating to the complexities of autonomous flight—the only viable means of executing the service, given the shortage and cost of human pilots.) So many billions were lost, congressional investigations became routine. But once urban air taxis finally materialized, going to the airport became so easy you might even forget you had to do it.

But Steven had his little voice—delivered through a pair of tiny form-fitting, haptic-technology cylinders that could stay in his ear canal all day, or be taken out in a second—to nudge him out the door. He picked up his coat and headed to the elevator, his suitcase following dutifully behind like a loyal puppy. In front of the building, he grabbed one of the motorized scooters known as “gliders” at the corner, waved his watch over the display to activate it, and tucked his bag into the tall storage compartment at the rear. The glider was much more comfortable than a regular scooter, with a wide footboard that allowed you to stand with your feet side by side, and an optional seat that extends from the storage compartment. It managed the light snow well, with its self-balancing, ice and snow traction assists, and small snow tires. You could race down a snowy street at full speed if you wanted—which Steven did.

He jumped into the lane right next to the sidewalk that was dedicated to gliders and the smaller scooters—there were also lanes for bikes and both autonomous and human-driven EVs; gasoline-powered vehicles had been banned entirely from the city for years—and hit the throttle, whirring off toward the air taxi terminal five blocks north. He arrived at the terminal building in just three minutes, since traffic was light and, as a driven vehicle instead of an automated one, he had the right of way. But the device wasn’t short on active safety features; he could have made the trip blindfolded, with the glider able to avoid close obstacles and all the other autonomous and connected vehicles able to steer clear of him from farther out. Stoplights were long gone in New York; getting around town was a breeze.

As was getting out of town, whether terrestrially or in flight. He placed the glider in a covered pen and walked over to the express elevator facing the sidewalk. He waved his watch at it and the door opened. Two minutes later, he was on the roof, 50 stories high, just eight minutes after his whisper device had reminded him to leave the office and well ahead of schedule. Steven walked past an upright scanner, which identified him and sent instructions to his whisper: “Pad 5. Join two other passengers who just arrived; all three of you are headed to the same terminal at JFK, 13 miles away.”

He walked out to the covered, wind-shielded walkway, admiring the flurry of a half-dozen e-VTOL aircraft arriving and departing in close succession. Snow fell steadily between them all, but was stirred up near the surface due to the persistent downwash from the aircraft. Steven pulled out his screen to scan some contracts while he waited. The seven-inch wisp of transparent glass displayed everything crisply and brightly, with all its processors embedded within a cylindrical handle no bigger than a drinking straw on one side. All three devices—the whisper, the watch, and the screen—were synced together as a system, one that had long since learned his preferences. The screen was the go-to device for observable media, but the whisper his primary interface; Steven gave voice commands, and the devices seamlessly obeyed. The trend had created a spooky effect: the sight of people walking down the street, randomly mumbling softly as they sent messages, requested information, and operated other connected devices. The AI-powered voice-recognition interface had become so natural and reliable, it quickly came to dominate the modern world, appearing in vehicles, cameras, and public devices like service robots, ticket counters, even restaurant menus. You simply whispered, and the devices would detect your facial muscle movements and isolate your voice to understand what you were saying. Steven gravitated to the system instinctively, just as he had to the air taxis. He was an early adopter to the bone.

As such, he’d watched the evolution of air taxis with palpable anticipation, grimacing with every misstep and failure. The system, as eventually deployed, was highly evolved—and regulated. Take the landing pads: Many buildings had their own spots on top for pickup, but since so many buildings had spires or other impediments to hovering air taxis, multi-pad rooftops had become the dominant arrangement among those that could host more than one craft. The idea of so-called “vertiports” with dozens of pads had been floated early on, but quickly abandoned as too expensive, too loud, and generally counterproductive to the cause. Nobody could stand the racket, and the swarming e-VTOL aircraft were an eyesore on the skyline, like gnats at a picnic.

Even with the distributed operations, city ordinances had limited the rollout of the air taxi service, primarily due to noise. The need to move high volumes of air in vertical flight was well ahead of the science intended to muffle those acoustics. It didn’t help that internal combustion engines had been banned from the city—significantly muting the city background noise that air taxis were intended to blend in with—and the urban canyons amplified the steady racket of e-VTOL aircraft, as the high-pitched electric whining bounced off buildings and into ears. As a result, New York City and most other urban areas limited air taxis to 4,000 flights per day, with a required minimum of three passengers on every multi-pad departure. This meant a fleet of about 200 aircraft across the city to start—a seeming pittance after 30 years of struggle.

That was just in the cities, though; the systems were up and running in hundreds of smaller metro areas and thousands of suburban and rural locales around the globe. Still, 200 aircraft in essentially constant use was a huge change from the turn of the century in New York, when spotting five or six helicopters in the airspace around New York at once was a challenge. Now you could catch dozens of VTOLs in every direction; that alone sparked vocal and ongoing pushback from the population, as did the limited broader benefit. Because of the movement restrictions, the system could only handle about 20,000 people per day at best, which barely put a dent in the city’s transportation needs. It helped that terrestrial transport had evolved concurrently with the rise of the air taxis, with congestion-battling autonomous vehicles, plenty of gliders and scooters, and a thoroughly modernized subway also on offer. Air taxis were thus seen more as a luxury for the wealthy, thanks to the $200-per-flight ticket cost. Though initially promised to be far cheaper, the movement restrictions forced operators, quite predictably, to increase their ticket pricing to pay for the expensive, fully-autonomous aircraft, which cost more than $700,000 each.

Uber's 2016 flying taxi concept, Uber

Steven used the service about five or six times per month, and only when his schedule demanded it. As routine as the flights had quickly become, however, he still enjoyed the experience—though it was still, frankly, a somewhat white-knuckle affair at times. The aircraft had been deemed safe, but they also had to be small and light in order to get enough range out of their battery packs, so they were built entirely out of pricey composite materials. The cabin walls were thin, and missing the layers of insulation and interior trim panels you might have on a car or a larger airplane. This was no flying Mercedes. It felt more like the old Toyota 86 Steven drove back in high school, though with a sleeker, modern interior—and, of course, wings and rotors. It also heated and cooled about as well as an old car; you learned to keep your coats and gloves on in the winter, and to crack the ventilation ports in summer.

As a result of their compact design and low weight, unpredictable low-altitude urban wind currents easily buffeted the taxis, though they moved confidently through still air while cruising higher. This vulnerability in dynamic, multi-obstacle environments—among the greatest challenges in rotorcraft autonomy over the years, and far harder than automotive autonomy or even fixed-wing flying—made it even more critical that they keep their distance from other air taxis and the buildings. The autonomous flight systems were thus told to coordinate their takeoffs and landings with other aircraft, get up above the city as quickly as possible on takeoff, and to land smoothly and efficiently while allowing as little time as possible for wind to upset the aircraft on the way down. As a result, departures and takeoffs alike were...sporty, to say the least.

Steven walked over to Pad 5 within a few seconds of arriving on the rooftop, and nodded politely to the other passengers, two women traveling together. A red light on a pedestal signaled for them to hang back for a few more minutes, until the aircraft finished charging and the pad was clear of moving aircraft. In the meantime, they each received their seat assignments via their whispers. The scanner they walked past on arrival had also calculated their respective weights, along with that of their luggage, and distributed the trio for the best balance. If it didn’t do that, the aircraft would be unstable and the range diminished; one or two of the motors would have to exert extra energy to compensate for the unequal weight distribution, and the aircraft would also be highly susceptible to loss of control during sudden evasive maneuvers. Steven was told to sit in #3—the right rear seat—while the two women took the front. Since there were only three of them, they were also asked to place their luggage in the fourth seat rather than the compartment behind the passenger cabin—again, to optimize balance.

The light went green and the three passengers trotted out in the snow, stashed their bags where instructed, and climbed into the aircraft. A young attendant walked over, confirmed everyone was buckled in properly, and secured the luggage with a tight net. He tapped a button to slide the expansive canopy forward and lock it securely in place, then he did a quick lap around the aircraft to check for ice, debris, damage, or any other trouble. Icing was a persistent concern in the winter, and the air taxis were equipped with heaters on the control surfaces, wings, and the rotor blades, in addition to having sensors that could detect ice based on the weather conditions and subtle changes in flight performance.

The vehicle had six rotors in total—four facing up and down, mounted at the ends of two sets of wings, and two facing forward and backwards, mounted on the tail assembly’s vertical stabilizer. The four wing rotors dominated during takeoff and landing, while the two tail rotors handled most of the forward momentum. Lift in forward flight came mostly from the wings, modestly supplemented by the wing rotors. The wings were pitched slightly upward at the leading edge; the aircraft flew canted slightly forward, to place the wings in a horizontal position and allow the wing rotors to generate both lift and forward thrust. The effect also made for great views for the passengers during the flight, as their seats were angled to be in a perfect position while cruising. Operators capitalized on that in their marketing, which often featured the aircraft whizzing past the new supertall skyscrapers towering above midtown Manhattan, the passengers pointing excitedly.

It was time to go, barely 12 minutes after Steven received the reminder in his ear. The attendant gave a thumbs-up to the passengers, who returned the gesture and put on their small noise-canceling headsets. A voice came on immediately: “Welcome aboard,” it chirped, in a gender-neutral-but-natural tone. (The system had debuted with a folksy Chuck Yeager drawl familiar to those who’d flown commercial aviation in previous decades, but it was immediately and mercilessly lambasted in social media.) “We’re ready to go, and will arrive at Terminal 2, JFK Airport, about eight minutes from now. We’ll land on roof pad number 14, directly above your gate areas. You have all been cleared for direct boarding, and your flights are on schedule in spite of the weather. Is everyone ready?”

All three said yes. “Terrific!” the autopilot responded. “Feel free to ask any questions about the flight along the way, and I’ll keep you updated about what’s going on.”

Within seconds, the rotors spun up to full throttle and the air taxi bounced straight into the air. It was always a startling sensation—the immediate g-force, the uncannily elevator-like ascent, and then, within seconds, the transition to forward flight at 3,000 feet, to keep clear of the taller skyscrapers. The computer, tracking all the buildings and other aircraft flight paths in the vicinity, plotted a course to JFK at precisely 120 mph—the informally-agreed-upon standard air taxi cruising speed, which made the math easy for the passengers: Ten miles took 5 minutes, 20 miles took 10, etc.

The taxi sailed across the city, the snow racing past the canopy like stars at hyperspeed. There were no swoops past skyscrapers, steeply banked turns, or other sci-fi nonsense; the flight was designed to be efficient, direct, and drama-free. The androgynous electronic narrator piped up regularly to let the passengers know where they were and what it was doing, pointing out other aircraft the system was tracking and structures that could pose obstacles if it had to divert for any reason. If the passengers wanted, they could turn off the narration—or they could double-down on it by donning lightweight goggles that used augmented reality to highlight other aircraft, show their own flight path, provide a more detailed stream of flight information, and point out interesting features around them.

They crossed the East River, then flew over Greenpoint, Bushwick, and the Evergreens Cemetery in a straight line, the neighborhoods emerging from the snow and scrolling slowly beneath them with clinical precision. The view was accompanied by the strains of a musical soundtrack composed specifically for these flights, a New Age elevator music-like tune with a slow tempo, deep bassline, and suspenseful strings. If you didn’t like it, of course, you could easily just divert your own playlist into your headset by whispering the list or specific tracks you wanted to hear.

Over Ozone Park, the airport finally came into view just as the air taxi started its descent. It took a hit from a gust in the wind, bouncing twice before settling. But a beat later, the passengers detected a moment of hesitation in the aircraft, followed by a sudden surge in power and a steep upward pitch.

The computer instantly told them what was happening: “Birds in vicinity—diverting!”

At that point, the years of persistence and commitment to the highest possible standards of artificial intelligence in autonomous engineering paid off. Vehicular autonomy isn’t about the act of flying or driving itself; going from A to B is easy, as are navigation and communication. Dumb autopilot systems from 50 years before could manage all that; old flying machines had sensors and scanners and control authority, and precise envelopes within which autopilots would allow the aircraft to operate. What they didn’t possess, however, was proper intelligence and judgment—the ability to interpret a scenario based on the available information and react to it. To see an airport van by the side of the runway, for instance, and have a hunch, so to speak, that it may not see the aircraft, or to see a newspaper blown up into the sky and know it wasn’t a drone or a bird. An airplane on autopilot keeps doing whatever it’s doing until greater intelligence intervenes, whether that intelligence is human or synthetic. AI-driven autonomy makes its greatest impact in the margins—the places where humans would previously intervene, assessing and deciding in a heartbeat what to do based on instinct, experience, and imagination.

That, of course, was the major hold-up that had made the road to autonomous flying machines so lengthy: developing such exquisitely advanced systems, infusing them with good judgment, and certifying them as safe for passenger flight. It all came from complex partnerships between software and aviation companies, coaxed along over decades. That extra time allowed the autonomous systems to rack up vast experience learning from human air taxi pilots—used in the early days, before autonomy.

When it comes to flocks of Canada geese flying near an airport, for instance, the taxis knew precisely what to do when a collision appeared imminent: accelerate and climb, since birds are known to dive when startled. And accelerate and climb is precisely what Steven’s air taxi did. 

However, one of the geese over Ozone Park didn’t do the expected. Instead, it bolted upward, directly into the path of the taxi’s front-right rotor, obliterating both itself and the rotor’s blades—and causing the motor and a chunk of wing to fall off. Ordinarily, that would have been fine; the remaining wings and rotors could have stabilized the aircraft even with one destroyed rotor. But a second vicious wind gust hit the taxi at the precise moment of the bird strike, something that may have been a factor in the bird’s own flight path, as well. Between that and the already-low airspeed that was a consequence of the initial descent into the airport and the subsequent attempt to climb out of the flock’s path, the taxi pitched over in the direction of the missing motor. It began to tumble.

Inside, the passengers could only wait, hope, and struggle to endure the intense, shifting g-forces. The autopilot’s voice, broadcasting through both the headsets and a cabin loudspeaker, turned instantly serious, instructing the passengers on the safety protocol: “ARMS CROSSED ACROSS CHEST, KNEES AND FEET TOGETHER; HEADS ON HEADRESTS.” In anticipation of a possible impact with the ground, their seatbelts automatically cinched up, and the airbags in the headrests primed for deployment.

The remaining rotors fought to correct the tumble, to little effect. But like a good pilot, the taxi also knew when all was lost—when it was time to give up. That moment arrived at 2,000 feet of altitude: The taxi waited until the aircraft was momentarily upright in its precipitous tumble and fired a small rocket out of the top of the cabin, just behind the canopy. That rocket pulled out a parachute that deployed instantly, with straps buried in the exterior skin peeling out of the surface and forming a self-leveling four-point harness. The chute inflated and the taxi stabilized beneath it, dropping gently toward the ground. The parachute had a steerable parafoil shape, and the remaining motors kicked back in to aim the taxi toward a safe landing spot, determined by laser scans of the ground and the system’s maps: a grassy circle inside a cloverleaf of the Belt Parkway, right outside the airport grounds. The ground came up fast, as the taxi steered itself toward the safe landing spot. The shock-absorbing landing gear deployed, and the craft touched down—hard, but safely.

Steven and his co-passengers were scared, but safe. They checked each other to confirm they were okay; one of the women was clutching her arm, which had been broken during the tumbling, but all three of them were otherwise all right. It was only the third air taxi accident in 10 years in the city, but the first responders' reaction to precisely these kinds of events had been well thought out: Some years before, New York City had introduced an emergency e-VTOL aircraft network across all five boroughs, with EMTs always within 100 feet of the aircraft. The ships were larger and faster than the taxis, able to carry two EMTs and two patients each in stretchers. Two were dispatched from the Queens location—a rooftop in Jamaica, just a mile or so away—the moment the air taxi’s parachute activated, and were themselves in the air even before the stricken aircraft hit the ground. The city’s automated air traffic control system diverted all other aircraft out of the ambulance’s path, so the duo of emergency vehicles reached the crash within three minutes, blue and red lights flashing urgently; they circled briefly to evaluate the scene, then landed next to the taxi.

As instructed by the autopilot, Steven and his fellow passengers had stayed in the vehicle, where it was safe: the emergency aircraft were landing, it was snowy and slippery out, and there was no immediate danger of fire. The EMTs helped them out and quickly loaded the woman with the broken arm into the air ambulance, her friend tagging along.

Steven, standing in the snow, wished them well, then looked back at the wrecked air taxi. The parafoil had draped itself over the rear tail assembly, and the mangled front right wing still had some goose feathers clinging to the strands of frayed carbon fiber.

“Do you want to go to the hospital, too?” the EMT asked Steven. “We recommend it, but you’re free to refuse treatment.”

“No, I’m fine—thanks,” he said, brushing snow off his jacket.

Just then, the taxi company’s official aircraft landed, with two representatives on board. They introduced themselves and asked if Steven was all right. He said he was. They asked him if he’d like to be taken back to the city via air or ground, or continue on to his destination. Steven shrugged his shoulders, and said he may as well continue. He’d only been on the ground 10 minutes or so, and the whole event had essentially proven nothing more than an aerial fender-bender, all things considered. (That said, news camera drones were already circling the area, automatically steering clear of the emergency vehicles.)

“Of course,” one rep said. “Our team will contact you for a statement and compensation for your trouble, and to make sure you’re okay. Would you like us to take you to the airport?”

“Sure,” Steven said, smiling at the weirdness of the whole experience—and the brevity.

He climbed into the representatives’ aircraft; one of them climbed in with him, while the other stayed behind at the accident site. The rotors spun up, and they took off. Less than a minute later, he was on the roof of his terminal at JFK—15 minutes late, but still with enough time for coffee.

As he stood in line for a cup of joe, he felt a slight thump in his ear, then heard a voice: “Your air taxi account has been credited five free flights. Your flight to San Francisco boards in seven minutes.”

stripe
Car TechSelf-Driving Tech