Reliable Robotics, the Cessna Caravan They Taught to Fly Itself, and the Retrofit Strategy That Sidesteps the Blank-Sheet Certification Nightmare

Reliable Robotics, a Mountain View, California startup founded in 2017, has demonstrated a fully autonomous Cessna 208B Grand Caravan - completing engine start, taxi, takeoff, climb, cruise, approach, landing, and engine shutdown with no one in the cockpit. Rather than designing a new aircraft from scratch, the company retrofits an existing, type-certificated platform and pursues approval through the FAA’s Supplemental Type Certificate process. That strategic choice may prove to be the smartest decision in autonomous aviation today.

Why the Caravan?

The Cessna 208B Grand Caravan is one of the most proven single-engine turboprops ever built. Introduced in 1984, powered by a Pratt & Whitney PT6 turboprop, and operated by FedEx Feeder, missionary aviation organizations, and floatplane services across Alaska, it has accumulated tens of millions of flight hours across four decades of service. Its safety record is documented in exhaustive detail and the FAA knows the platform intimately.

That established history is not incidental to Reliable Robotics’ strategy. It is the entire basis of it.

The Certification Problem Most Autonomous Aviation Companies Are Solving the Hard Way

The fundamental obstacle in autonomous aviation isn’t the technology - it’s the regulatory pathway. A brand-new aircraft requires a new type certificate from the FAA, which means proving every system, every failure mode, and every structural loading case from zero. Companies like Joby Aviation are in the middle of that process now. Historical averages for a Part 23 or Part 25 type certificate run eight to twelve years from design lock to first commercial delivery, at a cost of hundreds of millions of dollars before a single paying customer or package is on board.

Reliable Robotics asked a different question: find a platform that already carries a type certificate with decades of documented safety data, and add the autonomous systems on top of it.

How the Supplemental Type Certificate Strategy Works

The FAA’s Supplemental Type Certificate process - the STC - exists specifically for modifications to already-certified aircraft. It is not a shortcut. The agency requires rigorous proof that any modification doesn’t introduce new hazards or degrade existing safety margins, and the engineering depth required for autonomous flight controls means those conversations with FAA certification engineers will be extensive.

But the foundation is already built. The Caravan’s aerodynamic characteristics are proven. The PT6’s failure modes are documented in decades of service data. The airframe stress analysis exists. Reliable Robotics isn’t starting from zero - they’re building an automation layer on top of a platform the FAA already understands.

Co-founder Robert Rose came out of SpaceX, and the logic mirrors how SpaceX approaches hardware development: reuse what already works, build the innovation layer on a proven platform, and don’t rebuild what doesn’t need rebuilding.

What the Autonomous System Actually Does

The Reliable Robotics system replaces the Caravan’s mechanical flight control linkages with fly-by-wire actuators that accept commands from three sources: a pilot in the cockpit, a remote pilot at a ground control station, or the autonomous flight management computer. The aircraft can operate in any of those modes and transition between them.

The sensor architecture is layered: GPS for navigation, lidar and cameras for terminal-environment situational awareness, and standard pitot-static and inertial reference systems. A voting logic system cross-checks inputs across sensors, flags disagreements, and resolves conflicts conservatively - the same principle instrument pilots apply on an IFR approach, executed in milliseconds across a dozen data streams simultaneously. When sensors disagree, the system doesn’t guess. It flags the conflict and resolves it on the conservative side.

The approach and landing phase carries the hardest engineering work. Winds shift on final. Traffic conflicts emerge. The runway environment may not precisely match the database. The system has to absorb those variables and still put the airplane on the numbers.

The Mojave Demonstration

Reliable Robotics has demonstrated the full autonomous mission profile on the Caravan at Mojave Air and Space Port in California. Engine start. Taxi. Takeoff. Climb. Cruise. Approach. Landing. Engine shutdown. A full-size, type-certificated Cessna Caravan executed a complete flight autonomously with no one in the cockpit.

Demonstrated in flight and certified for commercial operations are separated by a substantial amount of regulatory work. But many companies in this space have never cleared even the demonstration bar.

The Remote Pilot in Command Model

Under Reliable Robotics’ proposed operational architecture, a licensed remote pilot sits at a ground control station with real-time visibility into the aircraft’s systems and the ability to take control at any moment. That remote pilot may monitor multiple aircraft simultaneously - not hand-flying each one, but serving as the trained authority who holds override capability for the autonomous system.

The principle of human oversight isn’t removed from this model. It’s repositioned. The automation executes; the remote pilot holds final authority. The parallel to air traffic control is instructive: a controller doesn’t fly the airplane, but holds authority over sequencing and separation.

Why Cargo Comes First

Cargo doesn’t have an opinion about whether there’s a pilot visible in the window. Nobody ships a package on a Caravan for reassurance - the payload needs to move reliably, on schedule, and at a cost that works for the operator’s economics.

That makes cargo a more tractable regulatory conversation than autonomous passenger transport. The political and liability landscape around flying people without a cockpit crew will take considerably longer to resolve. The FAA has shown it’s willing to engage on autonomous cargo operations in ways it isn’t yet ready to for passengers.

FedEx has signed an agreement with Reliable Robotics to explore autonomous Caravan operations for feeder cargo routes - the short-to-medium range, high-frequency, point-to-point runs that FedEx Feeder currently operates with pilots under Part 135. When the numbers include pilot costs, duty time limits, hotel and per diem, and the scheduling complexity of crewing a cargo run at two in the morning, the economic case for an autonomous system with remote oversight is not subtle.

Funding and Commercial Validation

Andreessen Horowitz led a Series B round that put approximately $35 million into Reliable Robotics. For a company not building a new airframe from scratch, that capital goes primarily toward engineering the automation system, navigating the certification process, and staffing the regulatory team - not tooling and manufacturing.

The FedEx relationship provides commercial validation that most aviation startups at this stage don’t have. There’s a real customer with real routes and genuine economic interest in making this work - a materially different position than many electric vertical takeoff and landing companies that raised large rounds in the 2020 to 2022 window and are now working through whether their technology can be commercially viable at any price the market will actually pay.

Realistic Timeline for Certified Operations

Three to five years to a commercial operating certificate for autonomous cargo runs is a realistic estimate, with significant regulatory milestones required along the way. The STC for autonomous flight controls on the Caravan is still in progress. The formal regulatory framework for remote pilot in command on commercial cargo flights is still being developed by the FAA. The data requirements for autonomous operations in actual instrument meteorological conditions are still being defined.

Anyone suggesting autonomous cargo Caravans are operating commercially within 12 to 18 months is being optimistic.

What the Retrofit Strategy Delivers Beyond Certification

The retrofit approach carries an advantage that’s easy to overlook: inherited operational infrastructure. The PT6 support network is established and global. Mechanics are trained on the engine. Parts supply chains are running. Fixed base operators have handled the Caravan for four decades.

When autonomous Caravan operations become certified, operators won’t be rebuilding their maintenance world around a new powerplant. They’ll be adding an automation layer to a platform they already know how to support. More than one promising aircraft program has learned the hard way that certifying the airplane was the easy part compared to supporting it in the field.

The Honest Limits of the Retrofit Approach

The Caravan is a 1984 design. Its aerodynamics, payload capacity, and cruise performance are fixed. A blank-sheet autonomous cargo aircraft could make different design choices - different wing loading, cruise altitude optimization, fuselage geometry purpose-built for freight. Reliable Robotics doesn’t have that freedom.

The propulsion question is also real. The PT6 burns Jet-A. The long-term trajectory for short-range cargo propulsion is likely moving toward hybrid or electric systems over the next one to two decades. An autonomous Caravan solves the automation problem but doesn’t address where propulsion is heading. Reliable Robotics has correctly prioritized making the autonomous cargo mission work commercially first - but the limitation is worth noting.

What This Means for Pilots

In the near term, the impact on pilot careers is limited. Certified autonomous cargo operations are still years away, and the remote pilot in command model keeps trained pilots in the loop - at a ground station rather than a cockpit, but still holding final authority.

In the longer term, if autonomous cargo becomes routine and remote pilots monitor multiple aircraft simultaneously, the nature of that work looks different from traditional flying careers. No instrument currency built in actual IMC. No logbook time accumulating on approaches. A remote monitoring role that may not translate to the kind of experience that advances a pilot up the career ladder in the traditional sense.

Whether that represents a door opening or a door closing depends on where automation goes over the next 20 years. Pilots who fly freight for a living should watch this company - not with alarm, but with the same clear attention you’d give a developing weather system before a flight into unfamiliar terrain.

Key Takeaways

Reliable Robotics has successfully flown a fully autonomous Cessna 208B Grand Caravan at Mojave Air and Space Port, completing the full mission profile with no one in the cockpit.
The company’s retrofit strategy - pursuing an STC rather than a new type certificate - leverages the Caravan’s existing FAA-recognized safety record and significantly compresses the regulatory timeline compared to clean-sheet programs.
A signed agreement with FedEx for feeder cargo routes provides commercial validation and a real economic case for the technology at a stage when most competitors have neither.
Andreessen Horowitz led a ~$35 million Series B; most of that capital goes toward certification and regulatory work, not airframe manufacturing.
Three to five years is a realistic estimate to certified commercial autonomous cargo operations - the regulatory frameworks for remote pilot in command and autonomous flight in IMC are still actively being developed by the FAA.