The aviation industry is beginning to explore a transition in which the aircraft itself, rather than a human sitting inside a cockpit, becomes responsible for much of the work required to complete a flight. While autonomous air taxis and electric vertical take-off and landing aircraft have attracted considerable attention, another technological race is developing more quietly around autonomous fixed-wing aircraft.

The first commercial applications are not focused on passengers. Instead, companies are targeting operations where removing the pilot can provide a direct economic or safety advantage, including agricultural spraying and cargo transportation. These early applications are effectively becoming real-world testing grounds for technologies that could eventually support autonomous passenger services.

One of the clearest examples comes from Pyka, a California-based start-up that develops fixed-wing aircraft without conventional cockpits. Its aircraft are designed for agricultural spraying and cargo operations, with the company ultimately envisioning larger fleets capable of transporting passengers.

At a crop-spraying test site in California's San Joaquin Valley, Pyka's aircraft can operate extremely close to the ground without exposing a pilot to the risks associated with low-altitude agricultural aviation. Russ Marotzke, a flight test engineer at the company, explains that the aircraft can fly lower than a human pilot would normally be able to.

The lower operating altitude has a practical economic and environmental consequence. By remaining closer to the crops, the aircraft can reduce spray drift, which means less agricultural chemical may be required to achieve the intended coverage. In this application, autonomy is therefore not simply replacing a person in the cockpit; it is changing how the underlying operation can be performed.

Pyka's aircraft are fully electric crop-spraying planes. Their batteries are positioned in the nose, while a central tank can hold as much as 300 litres of spray. The aircraft can remain airborne for approximately 35 minutes and have wingspans of around 11.5 metres, making the description of them as merely "large drones" somewhat misleading in terms of their physical scale and operational ambition.

At the test site, engineers define the area that needs to be treated using a computer interface. The software then generates the aircraft's flight path while accounting for previously mapped obstacles, including power lines and other hazards surrounding the agricultural area.

The aircraft can take off autonomously from a runway next to the field. During the demonstration, it flies for roughly 15 minutes before detecting that its supply of liquid is becoming low. It then returns to the ground and lands by itself. The aircraft is manually refilled and its battery is swapped during the demonstration, after which it takes off again and resumes spraying precisely from the location where it stopped.

This operating model illustrates the distinction between conventional autopilot and genuine autonomous flight. Autopilot systems generally assist a human pilot in controlling an aircraft, much as cruise control or lane-keeping technology assists a driver. An autonomous aircraft is intended to manage the complete flight process, including take-off and landing, by interpreting sensor information and using algorithms to control its actions with minimal or no direct human intervention.

The technology, however, has taken longer to mature than autonomous driving. Mykel Kochenderfer, a Stanford University expert in safe aviation autonomy, points out that major technology companies invested enormous resources into self-driving vehicles while comparatively less capital and attention went into autonomous aircraft.

There is also a structural reason for the slower progress. Aviation operates under significantly stricter safety expectations than road transportation. The potential consequences of an aircraft accident can be much more severe, creating a substantially higher threshold that autonomous systems must meet before regulators can authorize broad deployment.

Military applications have become an important accelerator for the sector. Several autonomous-aircraft companies have defence contracts that allow them to demonstrate and test their systems in military environments, where the regulatory pathway can be less restrictive than in civilian aviation. Some companies are already supplying military customers, allowing autonomous-flight technologies to mature through operational deployments before reaching wider commercial markets.

In the United States, Pyka's agricultural aircraft currently represents the largest autonomous fixed-wing aircraft approved for commercial civilian use. The authorization was obtained in the previous year, although the permitted operation remains tightly constrained. Flights must take place within a defined agricultural environment and require a ground operator as well as a visual observer. Pyka had previously obtained a comparable authorization in Brazil, where the regulatory environment is more permissive.

The company's next challenge is economic scale. Pyka currently produces roughly two dozen aircraft annually but aims to increase production to around 1,000 aircraft per year by 2030. Each aircraft sells for approximately $550,000, and customers receive training to operate the system.

That target demonstrates why autonomous aviation is ultimately as much a business-model question as a technology question. If production can be scaled and one trained operator can eventually supervise several aircraft, the economics of agricultural aviation could change substantially. The aircraft could potentially reduce labour requirements while increasing operational flexibility and allowing dangerous or repetitive work to be performed without placing pilots directly in harm's way.

Cargo is another major early market. British company Windracers is seeking permission to establish an autonomous cargo operation in Shetland and Orkney. Its aircraft are intended to transport goods to remote locations and have also been used for missions in Ukraine.

Windracers founder and chairman Stephen Wright describes the proposed operation as potentially the first heavy-lift air-cargo service conducted by drones in the United Kingdom and possibly the first of its kind more broadly. The commercial logic is particularly relevant for remote regions, where conventional logistics can be expensive, weather-dependent or difficult to operate efficiently.

Supporters of autonomous aircraft argue that these systems could address several structural problems facing aviation. A shortage of pilots could become less restrictive, workers could be removed from dangerous agricultural and logistics operations, aircraft utilization could potentially increase, and operating expenses could fall if a single person is eventually able to supervise multiple aircraft.

Automation is also presented as a potential safety mechanism. The argument is that aviation has historically become safer as increasingly sophisticated automated systems have been introduced, and that properly engineered autonomy could eventually continue that trend.

Pilot organizations, however, remain strongly cautious about removing humans from aircraft. The US Air Line Pilots Association describes the elimination of pilots as a serious safety gamble and argues that removing professional crews goes too far.

The concerns are particularly visible in agricultural aviation. The US National Agricultural Aviation Association has argued that small uncrewed aircraft can be difficult for pilots to see, while conventional crop-spraying aircraft operated by human pilots can cover considerably larger areas more quickly.

The industry is also divided over how autonomous aircraft should be designed. Pyka and Windracers are developing aircraft specifically for autonomous operation from the beginning. Their argument is that designing autonomy into the aircraft from the outset allows the structure, systems and operational characteristics of the plane to be optimized around its intended mission.

Other companies are taking the opposite route by modifying aircraft that already exist and have been certified for conventional operations.

US company Reliable Robotics, which has received investment from Boeing's investment arm, is testing autonomous technology on the Cessna 208B Grand Caravan. The aircraft is a single-pilot cargo plane capable of carrying approximately 1,360 kilograms of payload across hundreds of kilometres.

Reliable Robotics argues that retrofitting an already certified aircraft provides an important strategic advantage. Instead of having to prove the safety of both an entirely new aircraft and its autonomous technology simultaneously, the company can focus primarily on demonstrating that its automated flight system can operate safely on a known aircraft platform.

Merlin Labs is following another path. The US company has progressively tested its autonomy technology on increasingly larger military aircraft and is now applying the system to the Lockheed Martin C-130J military transport aircraft, which normally uses two pilots. The company's longer-term plan is to move toward commercial cargo aircraft designed for multiple crew members.

Merlin describes its system as a common "autonomy brain" that can move between different aircraft types. The strategy reflects a broader software-centric vision in which the intelligence controlling the aircraft becomes a reusable technological layer rather than a system permanently tied to one specific airframe.

The companies also disagree on the appropriate role of artificial intelligence. Reliable Robotics is deliberately avoiding AI in its autonomous flight system because the company believes that adding AI would make the certification process more complicated.

Merlin Labs, by contrast, is taking a much more AI-focused approach. This difference becomes particularly important in the development of detect-and-avoid systems, one of the central technical challenges facing autonomous aviation.

A human pilot can visually identify another aircraft, assess its trajectory and react to changing circumstances. An autonomous aircraft must reproduce that capability through sensors, software and decision-making systems, while operating with virtually no tolerance for serious mistakes.

There is no single perfect technical solution yet. Companies are therefore combining multiple sensor technologies and adding redundancy to existing systems in order to create additional layers of protection.

Reliable Robotics has installed forward-looking air-to-air radar capable of detecting other aircraft more than eight kilometres ahead. Its software then uses predetermined rules to determine how the aircraft should respond to the detected situation. The company argues that this system can provide detection capability beyond what a pilot's eyes can achieve in certain circumstances.

Merlin is taking a different approach by using AI-powered cameras capable of detecting and classifying objects. Instead of relying entirely on fixed rules, its system uses artificial intelligence to interpret what the cameras observe.

Pyka has used lidar technology since the beginning of its development process. Lidar allows its aircraft to detect objects including trees, vehicles, large birds and terrain. However, because lidar is relatively short-range, the company intends to supplement it with AI-powered cameras. This would represent Pyka's first significant use of AI onboard its aircraft.

Pyka's leadership sees AI as most useful in situations where conventional programmed rules become less effective. Distinguishing a distant aircraft from an unrelated visual mark or other object can require interpretation rather than simple measurement, making image-based AI a potentially valuable component of the safety architecture.

The question of artificial intelligence extends beyond avoiding physical obstacles. Autonomous aircraft operating in shared airspace must also communicate with air-traffic-control systems. Aircraft need to receive radio instructions, interpret them correctly and respond appropriately while operating within a highly regulated environment.

Reliable Robotics intends to keep a human component in this communication process. Its approach uses a remote pilot on the ground who is initially fully trained and responsible for communications and safety-critical decisions.

Merlin is pursuing a considerably more automated model. The company plans to use generative AI trained on thousands of hours of recorded air-traffic-control communications to interpret instructions and respond to controllers directly.

This represents one of the largest conceptual differences between the competing approaches. Reliable treats remote human supervision as an important safety layer, while Merlin views generative AI as a pathway toward reducing the human role in stages.

Merlin's proposed progression begins with aircraft operated by two pilots, moves to one pilot and eventually aims for a system requiring no onboard pilots. The strategy reflects the industry's broader attempt to transition gradually from familiar human-controlled aviation toward increasingly autonomous operations.

Pyka is taking a more cautious position regarding shared airspace. Rather than immediately attempting to establish its own solution to every regulatory and technical problem associated with autonomous communication, the company is willing to allow other developers to explore different operating models and identify the most effective approach.

The ultimate passenger market remains the most ambitious objective. Pyka co-founder and chief executive Michael Norcia describes a fully scaled passenger operation as the "holy grail", envisioning fleets of aircraft with minibus-like capacity moving passengers along the east and west coasts of the United States.

Norcia believes there is a reasonable possibility that autonomous fixed-wing aircraft could reach this stage before the electric vertical take-off and landing industry achieves a similarly large passenger network.

Whether that prediction proves correct remains uncertain. Passenger aviation introduces a substantially higher level of regulatory, technical and public-trust requirements than crop spraying or cargo delivery. An autonomous aircraft carrying people would have to demonstrate reliability across a far wider range of situations, including complex airspace interactions, unexpected obstacles, changing weather and emergency scenarios.

The more immediate significance of autonomous fixed-wing aircraft is therefore not that pilotless passenger flights are about to become routine. It is that the aviation industry is developing a new architecture for how aircraft can be designed, operated and economically deployed.

The competition is taking place across several layers simultaneously. Aircraft manufacturers are deciding whether to build autonomous planes from scratch or retrofit existing certified aircraft. Software developers are determining how much decision-making should depend on fixed rules versus artificial intelligence. Engineers are combining radar, lidar and cameras to create redundant perception systems. Operators are experimenting with different combinations of ground supervision, remote piloting and autonomous control.

These choices will influence the identity of the companies entering the market. Pyka is positioning itself around purpose-built autonomous aircraft and operational efficiency. Reliable Robotics is building its identity around safety-focused autonomy integrated with established aircraft platforms. Merlin is pursuing a software-driven, AI-centric model designed to make autonomy transferable between aircraft.

The economic opportunity is equally broad. Autonomous aviation could create new business models in agriculture, logistics, defence and eventually passenger transportation. The value proposition is not limited to eliminating pilot salaries. It also includes the possibility of increasing aircraft utilization, reducing exposure to hazardous work, improving access to remote communities and potentially redesigning entire logistics networks around aircraft that can operate with fewer human resources.

At the same time, the cost of failure remains the industry's defining constraint. Aviation certification is intentionally demanding because an autonomous system cannot simply be judged by how well it performs under normal conditions. It must demonstrate predictable and safe behaviour when sensors disagree, unexpected aircraft appear, communications become complicated or the operating environment changes.

For that reason, the future of autonomous aviation is likely to develop incrementally rather than through a single dramatic transition. Agricultural spraying and cargo operations provide controlled environments in which companies can accumulate operational experience, validate their systems and build regulatory confidence.

Even if fully autonomous passenger aviation remains distant, the technology being developed today could still influence conventional commercial flying. Autonomous systems may first become additional safety and decision-support layers before they are trusted to assume complete responsibility for flight.

That intermediate stage could ultimately prove just as important as the final ambition of pilotless aircraft. Rather than immediately replacing pilots, autonomy may first redefine what pilots do, moving human expertise away from continuous manual control and toward supervision, exception handling and high-level decision-making.

The development of self-flying aircraft therefore represents more than another chapter in the automation of transportation. It is a strategic redesign of the aircraft as a product, the pilot as a role, the operator as a business model and the aviation company as a technology brand. The most successful players may not simply be those with the most advanced aircraft, but those capable of combining certification, reliable sensing, intelligent software, scalable production and commercially viable operations.

In that sense, the race for autonomous aviation is already underway. Crop fields and cargo routes are becoming the testing grounds for a technology whose eventual destination could be a much broader transformation of the skies.