Reliable Robotics and the autonomous Cessna Caravan flying cargo without a pilot on board

Reliable Robotics has flown a Cessna 208 Caravan — a 6,500-pound turboprop — with no one in the cockpit. The aircraft taxied, took off, flew its route, landed, and returned to the ramp entirely under remote command. This isn’t a concept or a simulation. It’s a real, full-size airplane operated by a pilot sitting at a ground workstation, potentially hundreds of miles away.

Who Is Reliable Robotics?

Founded in 2017 in Mountain View, California, by Robert Rose and Juerg Frefel — both former SpaceX engineers — Reliable Robotics takes a fundamentally different approach from most autonomous aviation companies. Instead of designing a new airframe, they retrofit the Caravan, one of the most proven utility aircraft in history.

The Caravan has been hauling cargo, passengers, and skydivers across the worst strips in the world for nearly 40 years. FedEx operates hundreds of them. They fly into unpaved runways in Alaska and land on floats in the bush. The airframe isn’t the question mark. It never was.

How Does the Technology Work?

Reliable installs a full fly-by-wire retrofit kit that replaces mechanical control linkages with electromechanical actuators. The system includes:

Redundant flight computers
Redundant GPS and communications links
Radar altimeters
A full sensor suite for environmental perception

The result is a Caravan that can be commanded from the ground by a remote pilot in command — someone who holds an ATP certificate and remains responsible for the flight. This concept is called remotely operated aircraft, distinct from fully autonomous flight because a human is always in the loop. The remote pilot can see what the airplane sees, intervene at any time, and communicate with air traffic control.

In 2020, Reliable completed an uncrewed flight of the Caravan at a test range — no safety pilot on board. The FAA has granted Reliable a special airworthiness certificate for their modified Caravan, signaling that the agency has reviewed the system and is willing to support controlled testing.

Why Start With Cargo?

The initial target is Part 135 cargo operations — specifically the feeder routes that FedEx and UPS rely on every night. Thousands of single-pilot Caravan flights move packages between distribution centers and smaller airports on short hops, often under 200 miles.

The pilot shortage hits this sector hard. These are night flights, short legs, repetitive routes — not glamorous work. Recruiting and retaining pilots for cargo feeder operations is one of the industry’s biggest headaches.

Reliable’s value proposition is straightforward: if a remote operator can supervise multiple aircraft simultaneously — perhaps three or four — the economics change completely. Ground operators can also hand off to fresh operators without the airplane ever landing, eliminating the fatigue problem and decoupling scheduling from crew duty times.

Cargo also offers a simpler path to public acceptance. An empty cockpit flying FedEx boxes at 2 a.m. over rural Kansas is a very different conversation than one flying passengers over a metropolitan area.

What Are the Biggest Technical Challenges?

Communications latency is the first concern. Commands from a remote pilot travel over a data link, and that round trip takes time. During normal cruise, a few hundred milliseconds is fine. During a gusty crosswind landing, it could matter. Reliable addresses this with onboard automation that manages the short-term flight path while the remote pilot provides high-level commands — more like managing an autopilot than hand-flying.

Lost link scenarios present the second challenge. If the data connection drops, the airplane follows a pre-programmed contingency management system — holding, diverting, or continuing to its destination depending on flight phase and remaining fuel. Regulators want exhaustive testing of every failure mode, and that testing takes years.

National airspace integration is the third hurdle. Controllers currently assume a pilot is listening on frequency. Remote operations require reliable voice relay or new communication protocols so ground operators can talk to approach, tower, and center. Reliable has demonstrated this in test flights, but scaling to hundreds of nightly flights across dozens of facilities is a different problem.

Public perception remains the fourth challenge, and it’s the reason Reliable is starting with cargo rather than passengers. Proving the safety case with thousands of cargo flights builds the data foundation for everything that follows.

How Does Reliable Compare to Competitors?

Xwing, another Bay Area company, pursued a similar Caravan-based approach and completed an autonomous gate-to-gate flight. However, Xwing merged with Daedalean, a Swiss computer vision company, and shifted toward selling autonomy software rather than operating aircraft. This leaves Reliable as arguably the most focused company pursuing end-to-end remotely operated cargo with a full-size airplane.

Merlin Labs in Boston works on autonomy for military and larger cargo aircraft using more AI and machine learning in the flight management system. Reliable leans more heavily on deterministic automation with the remote pilot providing decision-making. The deterministic approach is easier to certify because every branch of the logic tree can be tested. The AI approach is potentially more capable but much harder to validate to FAA standards.

Most eVTOL companies — Wisk, Joby, and others — are asking regulators to trust a brand-new aircraft design and autonomy simultaneously. Reliable asks a simpler question: what if the airplane is already proven and you only change who flies it?

What’s the Realistic Timeline?

Reliable has targeted initial commercial operations in the mid-2020s, but they do not yet have operational approval. Aviation certification timelines consistently slip. A realistic estimate for the first revenue cargo flights under FAA approval is late 2026 or 2027, and that’s optimistic. Full-scale fleet conversion is a late-decade story at the earliest.

The worldwide Caravan fleet numbers around 3,000 aircraft, with a significant portion flying cargo. If even a fraction of those routes qualify for conversion, it represents hundreds of aircraft. Each conversion requires the fly-by-wire kit, sensor suite, and ground station infrastructure. While per-unit costs haven’t been published, pilot compensation, training, benefits, and scheduling overhead can run well into six figures per year per aircraft — the math likely works even at a high conversion cost.

What Does This Mean for General Aviation?

The broader trend is already visible. Autoland systems exist. Autothrottle is becoming common in turboprops. Emergency descent modes, automatic go-arounds, and envelope protection are increasingly standard. The cockpit is getting more autonomous whether the industry calls it that or not.

Reliable Robotics is taking the next logical step — asking what happens when you remove the pilot seat entirely. That’s not coming for light GA aircraft anytime soon. But the engineering is closer than most pilots have assumed.

Key Takeaways

Reliable Robotics has flown a full-size Cessna Caravan with no pilot on board, completing taxi, takeoff, route flight, landing, and taxi-back entirely via remote command.
The company retrofits proven airframes rather than building new ones, simplifying the certification path and leveraging the Caravan’s 40-year safety record.
Part 135 cargo feeder routes are the initial target, where pilot shortages are acute and public acceptance barriers are lowest.
A remote pilot in command is always in the loop, making this remotely operated rather than fully autonomous — a critical distinction for FAA certification.
Realistic first revenue flights are likely in late 2026 or 2027, with fleet-scale conversion still years beyond that.