Jump Aero will collaborate with Caltech to research an adaptive flight controller that could help eVTOL aircraft land safely after sustaining major damage, such as from a bird strike.
The Petaluma, California-based eVTOL developer will work with Professor Soon-Jo Chung’s autonomous systems controls laboratory to study the technology, which they’ve dubbed an “electronic parachute” for its potential to stop aircraft from plummeting out of the sky after a catastrophic failure. (No actual parachutes are involved.)
The research is being funded under a Phase I Small Business Technology Transfer (STTR) contract from the U.S. Air Force’s Agility Prime program, which aims to accelerate development of the commercial eVTOL industry. According to Jump Aero founder and president Carl Dietrich, the project will combine Caltech’s “bleeding edge research” on autonomous controls with Jump Aero’s practical approach to aircraft design and certification.
“In the controls world, there are amazing things that the researchers and the professionals out there can do, but the vast majority of this new stuff is just not practical to certify [for aviation],” Dietrich told eVTOL.com.
But if these complex control routines are not currently certifiable for baseline flight controllers, they could potentially win regulators’ approval in the form of an emergency recovery function, he suggested. He pointed to Garmin’s Emergency Autoland system — which can bring general aviation aircraft in to land at the nearest suitable airport in the event of pilot incapacitation — as an example of “the best practical way to bring some of these more complex control technologies to market.”
Dietrich and Chung — who previously worked in the same laboratory at MIT while completing their PhDs — envision using simulation and machine learning to develop alternate control laws for hundreds or even thousands of potential failure scenarios that would be impractical or impossible to test in real life. These would then be ported over to the actual aircraft, but remain dormant until the pilot activated an emergency switch.
The adaptive controller would then automatically assess the state of the aircraft and select the pre-learned routine that provided the best control authority for the available functioning effectors.
“Especially with a failed airframe, you can’t guarantee safe outcomes, but this basically puts all the possible technology tools at the disposal of a pilot in terms of what modern control theory can do to maximize the probability of a safe landing,” Dietrich said.
He emphasized that the collaboration is “true research . . . we don’t know if this is going to work.” Consequently, Jump Aero is not counting on incorporating the technology into its JA1 eVTOL, which it is developing for use by first responders.
However, Dietrich continued, “it’s certainly a technology where if it’s proven out to work and have this kind of benefit, then I could imagine this type of recovery function being implemented on any aircraft that has a fly-by-wire control system.”
The Phase I research will be conducted entirely in the JA1’s flight simulator. Various failures will be simulated; for each, the performance of the baseline controller will be compared to the performance of the adaptive machine learning based controller. If the program is funded for Phase II, Jump Aero and Caltech will progress to testing using subscale demonstrators.
The contract is the second U.S. Air Force STTR award for Jump Aero, which last month announced another Phase I contract for a partnership with the Unmanned Systems Research Institute at Oklahoma State University (OSU). That project will explore the possibility of using Jump Aero’s JA1U unmanned technology demonstrator for the calibration of navigational aids and instrument landing systems (ILS) at airports.
Currently, ILS calibration in the United States is performed by the Federal Aviation Administration (FAA) using a Bombardier Challenger jet. OSU researchers have developed compact technology that can potentially accomplish the task more cost-effectively using small consumer drones, but flying a drone on final approach would typically require runway closure.
Jump Aero thinks its high-performance JA1U demonstrator will be capable of integrating more seamlessly with general aviation (GA) traffic, potentially enabling it to perform the calibration mission with less disruption to airport operations. Dietrich said the JA1U will weigh just under the 55-pound (25-kilogram) weight limit for drone operations under the FAA’s part 107 regulations, but should be capable of flying well in excess of 90 knots (over 165 kilometers per hour) — typical speeds for small GA aircraft on approach.
This, too, is early-stage research with no guarantee of success. Nevertheless, Dietrich said, “it could conceivably progress to something that’s actually a fieldable system in the future, where we’re providing cost savings to the government because they don’t have to fly around a large Challenger airplane . . . but doing it in a way that is less intrusive than drone-based calibration would be today using a multicopter.”