Aurora Eyes Future of Robotic Copilot Technologies

Aurora Flight Sciences, which recently wrapped up a government-funded phase of its Aircrew Labor In-Cockpit Automation System (Alias) research, is evaluating a number of options for future robotic copilot research and potential applications for the resulting technologies.

The Manassas, Virginia-based advanced aerospace technologies specialist was among the contractors selected by the Defense Advanced Research Projects Agency to develop and flight-test Alias, which can act as a pilot assistant and operate an aircraft from takeoff to landing.

The formal DARPA program funding was complete in late summer, said Jessica Duda, Alias program manager for Aurora. But the project and the associated technologies have gained “broad interest” from both the military and commercial sectors, Duda said, adding the program demonstrated a number of key elements that show promise for future applications.

“We were able to install the system or portions of these systems on five different aircraft and we flew it on three of them. That was a big accomplishment for us,” she said. Aurora’s Alias research involved testing various aspects of the system on a Boeing 737-800NG simulator, along with Diamond DA42, Cessna 208 Caravan, Bell UH-1 Iroquois and de Havilland DHC-2 Beaver aircraft. The system flew on the DA42, Caravan and UH-1.

Duda highlighted several of the key elements demonstrated in the program, including the ability to develop a non-invasive automation system. New aircraft have increasingly more automation, which is built into the latest avionics systems. But legacy aircraft often can’t accommodate such systems, she said. “In those cases there is a distinct advantage to applying aftermarket automation into these aircraft to provide a pilot on board more assistance and to eventually reduce required crew.”

She added that the use of five different aircraft demonstrated the ability to rapidly transfer various aspects of the automation system. Aurora took a toolkit approach, developing a core system that could adapt to different aircraft and aircraft systems, from variations in each aircraft’s systems and handling qualities to how controls are actuated and how a pilot/ground operator interacts with the system.

The Aurora system involved substantial knowledge acquisition, she said, noting the company developed a “digital type rating” for each aircraft. But it also involved more practical accommodations, including fitting the hardware in various cockpits. The goal is to bring that adaptation time down to 30 days. “We were doing quite well with that,” she added. “We were able to swap out various systems as we moved from aircraft to aircraft." Having the ability to adapt the system was essential because “one size does not fit all.”

The tests varied on each aircraft, from use of a robotic arm to actuate the controls of the Boeing 737 simulator to the monitoring system on the Beaver to a full robot on board the Caravan.

She noted different levels of interest that had surfaced as a result of the programs, from the knowledge acquisition side to the development of robotic arms to move controls.

While the robotic arm offers promise of a second pair of hands, another key technology developed through the research involved a second set of eyes. Aurora developed a system involving a series of cameras—a machine vision system—that were used to watch all of the gauges, knobs and switches and other systems. The visual system was able to synthesize the information viewed in the cockpit, right down to the position of levers and switches and angle of knobs. That information could be used to provide alerts on any number of things, Duda said.

One of the more interesting aspects of the program, she added, was the interaction involving the pilot. “This is a really interesting problem to look at when you place a highly capable automated system literally next to a pilot on an aircraft,” she said. “How do you best communicate information between the pilot and an automated system and how do you best allocate [workload]?”

Aurora worked with a substantial number of operators to obtain feedback in this area. The approach, Duda said, was to give the pilot control over the workload. The pilots use a tablet to interface with the automated technology. “We’re not telling pilots the robot is going to do everything and pilots should sit and twiddle their thumbs,” she said. Rather, the pilot has the discretion over workload.

This research, along with each of the technologies, has captured various levels of interest, and Aurora is eying multiple potential paths, Duda said.

She believes the individual technologies may see their way to market sooner than a whole stand-alone robot, which is a much bigger commitment and cultural change.