Pioneering the Future of Aviation Safety

Pioneering the Future of Aviation Safety

Pioneering the Future of Aviation Safety

An Airbus innovation center uses a King Air C90 to collect data that drives its approach to incorporating rapid technological advancements in commercial aviation.

Acubed, pronounced A-cubed, is Airbus’ Silicon Valley innovation center that specializes in artificial intelligence and machine learning applications for aviation. The organization is actively conducting flight testing in a specially modified 1974 King Air C90 (serial number LJ-615) with the goal of using collected data to develop a variety of applications that enhance safety and efficiency for commercial aircraft.

Operating a flight lab

The King Air is being used to bolster a repository of advanced data that Acubed’s flight test team has collected over several years.

The flight test team consists of eight engineers and one test pilot.

“Our specially modified King Air represents the perfect blend of operational flexibility and technical sophistication – equipped with custom nose and tail camera mounts that replicate commercial airliner vantage points, enabling us to collect data that mirrors real-world A320 operating conditions,” a company social media post explained. “This aircraft builds on our successful Baron Flight Lab program and the platform gives us the size, payload and power to integrate more advanced sensors and accelerate our research toward safer, more capable flight operations.”

The King Air provides the Flight Lab team convenient and efficient access to flight testing and the ability to quickly test and mature new concepts.

Artificial intelligence in aviation

Paul Smith, director of flight test and operations for Acubed, noted that the innovation center based in Sunnyvale, California, is wholly focused on accelerating artificial intelligence and autonomy adoption in aviation, enabling Airbus to capitalize on rapid technological advancements.

“What we are really looking at is how to apply AI and machine learning,” he said. “AI is a huge, broad term, but I don’t like the term in aviation as it implies intelligent reasoning. So, we are looking at how we can take machine learning technology and inject it into Airbus’ different systems. This includes how we can use computer- and vision-based techniques for improving landing and obstacle detection.”

The proposed technology will not include new infrastructure outside of an aircraft, rather a self-supporting system that analyzes millions of data points to help pilots make sound decisions during critical phases of flight. To date, the team has conducted more than 1,800 approaches and collected almost 23 million images as a baseline repository for the complex machine to internalize and analyze.

How Acubed collects data

Acubed’s flight data collection efforts were initially performed using a Beechcraft Baron. The aircraft was flown in support of the Wayfinder project, which focuses on certifiable autonomous flight and machine learning solutions for Airbus, and collected most of the initial images and videos that are being used as the base for the application.

Acubed has collected nearly 23 million images with advanced cameras on the Beechcraft Baron and King Air flight test aircraft. Initial flight envelope testing in the King Air began in January 2025 and validation of the camera and software system followed shortly thereafter.

“We have been doing this project for about two and a half years,” Smith said. “We are concentrated on the airports that are serviced by Airbus A320 series and have flown to nearly 200 airports in the United States. We have flown all the way from San Diego, California, to Portland, Maine, and Miami, Florida, to Seattle, Washington, and all the Class Bravo airports between.”

Smith explained that aircraft today, if not flying a visual approach, will rely on an approach system to help guide them to the runway safely. Whether it’s an ILS, GPS RNAV or another approach, these existing, standard aviation systems help guide pilots safely to the ground.

“[These tools] help you have a stabilized approach and end up with a successful landing, so we feel like having a camera and a computer in the airplane that can monitor that approach and provide guidance back to the pilot is a safe thing to do And it’s autonomous in the sense that it doesn’t require anything on the airfield and adds redundancy to the pilots’ eyes,” he said, noting that there will also be an obstacle detection component that will alert crews to threats on the runway.

An ideal flight test platform

Initial image gathering efforts were conducted when there was a lone camera positioned in the cockpit. There was a lot of distortion when the lens was directed at the plexiglass canopy, so the need to formally affix the camera was apparent.

“We modified our 1974 King Air C90 with seven cameras in the nose and three in the tail. Those are arranged in a field of view that allows us the optimum view of the runway and airfield environment,” Smith said.

The King Air’s size translates well to where the technology will ultimately be used.

“The cameras we mount in the tail, which are different from those used on the Baron, are due to the fact we wanted the perspective of where the camera would be mounted on a typical narrow-body aircraft – about 14 feet off the ground,” he said. “If you think about taxiing around, having that height and relative angle of view provides a lot more fidelity in how we are trying to locate and determine the position of those objects when taxiing.”

The Baron is a good platform for this type of work, but there are several distinct advantages to its larger Beechcraft family member. The flight test team consists of eight highly educated and experienced engineers, as well as one test pilot. The King Air allows more of the Acubed team to be aboard and actively involved during test missions.

“One of the reasons we moved from the Baron to the King Air was a larger cabin,” Smith said. “This allows us the ability to do real-time development work in the airplane during flight, because we can put an engineer in the back who can operate, monitor and change the software of the system that we are using. We have a high-accuracy inertial unit that we use to provide location information for the cameras that’s separate from the avionics system and the King Air itself.”

Having an engineer in the back is particularly useful, as the team can correct course mid-flight if any revisions are needed. The speed of the twin turboprop is also a big step up from the twin piston. Additionally, the King Air fits in busier terminal environments better than the Baron did.

“When you’re operating in the Bravo environment, you want to be able to fly at the same airspeed as other commercial traffic,” Smith said. “So, that was one of the main reasons we went with the larger King Air. Also, getting to the airport is important, even though we have already flown to most of them [that we plan to] in the Baron. The King Air is going to be used to collect specialized information that we need to update the system. The King Air allows us to get there fast at a good altitude and interface more with commercial traffic.”

Smith highlighted that the C90’s higher cruising altitude allows the team to collect approach data starting at a higher altitude than it did in the Baron.

Major modifications required

Before adding it to the program, the C90 was overhauled from tip to tail. The intensive work included landing gear and engine touchpoints, as well as a new inertial navigation system and server rack with improved electronics and cooling. The aircraft’s flight deck was modernized with a full Garmin system that includes two Garmin GTN 750Xi displays, a G600 TXi, a TXi EIS monitor and a GFC 600 digital autopilot.

The camera system was the team’s most important consideration, and much work went into ensuring its design was appropriate for the flight test regime’s desired outcome. A custom nose mount was built, which houses a modular camera system and navigation sensors. The system has 10 cameras that collect synchronized images throughout the test flights. The nose and tail sensors’ wide-angle lenses are important in collecting data that is expected to be used for an obstacle detect-and-avoid system.

The 14-foot-tall tail of the C90 lends itself well as an ultimate transition to the A320. There were still certain attributes that needed to be considered when determining whether the King Air would be an appropriate application. The engineering team had to consider all elements of the aircraft’s design, focusing specifically on factors such as aerodynamic loading, thermodynamic effects, RF signal attenuation and more.

Acubed has collected nearly 23 million images with advanced cameras on the Beechcraft Baron and King Air flight test aircraft. Initial flight envelope testing in the King Air began in January 2025 and validation of the camera and software system followed shortly thereafter.

More opportunity to connect aviation + technology

Smith has been involved in flight testing throughout his career, noting that the work is never really done. He said Acubed will likely fly the King Air for at least the next 10 years.

“We are looking at quantum magnetic navigation and other systems that probably are in the future that we don’t even know about yet and that would be able to be fitted on this aircraft,” he said. “The King Air is a very efficient airframe for collecting and obtaining data versus a larger Airbus A320 or A350.”

The platform will also be used for continuous testing and needed refinements to other projects the Wayfinder team is working on, even after commercialization of the initial technology. Smith is excited by this work and invites others to explore the intersection of aviation and artificial intelligence.

If you were one of the approximately 704,000 visitors at EAA AirVenture Oshkosh this year, you may have seen Acubed’s King Air C90 sitting in front of the Airbus pavilion.

“A benefit of our presence at Oshkosh was reaching out to young folks and getting them more interested in aviation,” he said. “What I would emphasize is that if you’re into computers, if you’re into virtual reality, augmented reality, artificial intelligence or machine learning, don’t discount the aviation field as a very ripe area to expand.”

About the Author