A robotic copilot has landed a simulated Boeing 737 in a test to demonstrate how advanced automation could overcome pilot incapacitation.
The exercise was conducted by Aurora Flight Sciences under DARPA’s ALIAS (Aircrew Labor In-cockpit Automation System) program to develop a drop-in kit that adds high-level automation to existing aircraft.
The tests were performed in a 737-800 flight simulator at the U.S. Transportation Department’s Volpe National Transportation Systems Center in Cambridge, Massachusetts. The right pilot’s seat was replaced with the robotic arm and machine vision system developed under ALIAS.
The goal was to show the robotic copilot could use what was in the 737 cockpit, principally autopilot and auto-land, without having to integrate a new flight control system to enable the automation, said Jessica Duda, Aurora’s ALIAS program manager.
Aurora has previously flown ALIAS components on a Diamond DA42 light twin; a Cessna Caravan turboprop single; a de Havilland Canada DC-2 Beaver; and tested them in a Bell UH-1 helicopter.
The 737 simulator trails used the same tablet to enable the pilot to interface with ALIAS, but added speech interaction. This allowed the pilot to verbally command ALIAS, for example telling the automation to change attitude or heading. The system was programmed with the 737’s emergency procedures and the robotic arm flew the aircraft by manipulating existing cockpit knobs and switches. A camera was added to the end of the arm so the system could confirm the correct action had been performed.
In the simulated pilot incapacitation scenario, ALIAS initiated an autopilot procedure to descend to 10,000 ft., where the pilot could recover from hypoxia, and programmed the flight management system, via its keyboard, to circle a point until the pilot made an input. When the pilot did not make an input, ALIAS initiated an auto-land, extending the flaps and landing gear and commanding reverse thrust after touchdown. “It all went as programmed,” Duda said, adding that the 737 trials demonstrated that ALIAS can be rapidly programmed.
In addition to the tablet interface, voice interaction, robotic arm and a “knowledge acquisition system” to program in all the procedures from the aircraft flight manual, Aurora’s ALIAS includes a vision system that reads the cockpit instruments and monitors switch positions. This enables the automation system to fly the aircraft even if it lacks an avionics data bus and autopilot. In an emergency, the system will display the checklist, divide the tasks between pilot and automation and monitor to ensure they are all performed correctly.
DARPA’s goal is to develop a removable kit that can be adapted rapidly and installed quickly to enable reduced-crew operations in any existing aircraft. Aurora and Sikorsky conduced ALIAS Phase 2 flight demonstrations in late 2016, after which Sikorsky was selected to move into Phase 3. Aurora’s DARPA-funded work will wrap up shortly, but, “We have the intention to move forward with parts of the integrated system on a variety of aircraft,” said Duda.
The main test aircraft has been Aurora’s DA42-based Centaur optionally piloted aircraft. In the most recent flight demo, ALIAS executed cockpit procedures in real time and, overseen by an onboard safety pilot, conducted a fully automated landing on a simulated runway at 3,000 ft. altitude.
“ALIAS has proven its versatile automated flight capabilities,” said John Wissler, vice president of research and development. “As we move towards fully automated flight from takeoff to landing, we can reliably say that we have developed an automation system that enables significant reduction of crew workload.”