In a scene straight out of a sci-fi movie, a crop-spraying plane skims over an alfalfa field in California's San Joaquin Valley - except there's no pilot on board. The aircraft, operated by startup Pyka, flies so low that it's practically tickling the crops. 'We can actually go lower than a human pilot can,' says Russ Marotzke, a flight test engineer at Pyka, as the plane buzzes the field. That low altitude means less spray drift and fewer chemicals needed, a win-win for the environment and the company's bottom line.
Pyka, based in a converted WWII hangar overlooking San Francisco Bay, builds self-flying aircraft without cockpits, designed for crop spraying and cargo delivery. It's among a handful of companies racing to bring autonomous fixed-wing planes into commercial service. While urban air taxis (eVTOLs) hog the spotlight, a quieter race is underway for self-flying planes, starting with jobs like crop dusting and package delivery - and eventually, maybe, passengers.
'Fully scaled, ubiquitous passenger operation is the holy grail,' says Michael Norcia, Pyka's co-founder and CEO, who dreams of a fleet of minibus-sized planes ferrying people along the US coasts. 'There's a decent chance we'll get to that point before the eVTOL industry.'
At a test site about 80 km east of the factory, Marotzke and a colleague are testing a software update. The planes are fully electric, with a battery in the nose, a 35-minute flight time, and a 300-liter spray tank. They have an 11.5-meter wingspan - calling them 'large drones' feels like calling a whale a big fish. Engineers map the spray area on a computer, and the software plans a route avoiding obstacles like power lines. The takeoff is seamless, and after sensing low spray, the plane lands itself for a refill and battery swap, then resumes precisely where it left off.
Autonomous flight differs from autopilot. Autopilot assists, like cruise control in a car, while autonomous systems handle everything, including takeoff and landing, using algorithms and sensor data. Self-flying planes have been slower to emerge than self-driving cars, partly because tech companies poured billions into cars, says Mykel Kochenderfer, an expert at Stanford. But also because aircraft safety standards are stricter. 'The consequences for air accidents can just be so severe,' he notes.
Military interest is propelling the technology, with many companies holding defense contracts. In the US, Pyka's crop sprayer is the largest autonomous fixed-wing aircraft approved for commercial civilian use, but operations are limited to agricultural settings and require a ground operator and visual observer. Brazil is more permissive, and about a dozen Pyka planes are already spraying cotton and soybeans there. Pyka aims to scale production from two dozen planes a year to 1,000 by 2030, each selling for $550,000.
The UK hasn't approved such operations yet, but British firm Windracers is seeking permission for an autonomous cargo service in Shetland and Orkney, and its aircraft are also flying missions in Ukraine. 'It would be the first heavy-lift air cargo service by drone certainly in the UK and probably anywhere,' says Stephen Wright, Windracers' founder.
Supporters tout benefits like addressing pilot shortages, removing people from dangerous work, improving efficiency, and cutting costs. They argue automation could make flying safer, citing past declines in accidents with more automation. But the US Air Line Pilots Association (ALPA) calls removing pilots 'a serious gamble with safety and a step too far.' The National Agricultural Aviation Association notes small uncrewed aircraft can be hard to see and that piloted planes can spray larger areas faster.
Startups differ in approach: Pyka and Windracers build planes from scratch, while others retrofit existing aircraft. Reliable Robotics, backed by Boeing, is testing its system on the Cessna 208B Grand Caravan, a single-pilot cargo plane. Merlin Labs is working up to the Lockheed Martin C-130J military transport, with commercial planes next. 'It is a common autonomy brain that can transition between different aircraft,' says Merlin's CEO Matt George.
They also differ on AI. Reliable is avoiding it for certification reasons, using radar and fixed rules for detect-and-avoid. Merlin uses AI-powered cameras and generative AI to interpret air traffic control instructions. Pyka uses lidar but plans to add AI cameras. 'For figuring out that six pixels in the distance are an airplane versus some other smudge, it is perfect territory,' says Norcia.
Communication with air traffic control is a challenge. Reliable uses a remote pilot on the ground, while Merlin plans to use generative AI trained on thousands of hours of recorded exchanges. Pyka is content to let others 'blaze the trail.' Even if fully autonomous passenger flight remains elusive, the technology will likely inch into commercial aviation, making piloted flying safer - a welcome development, even ALPA agrees.