On September 3, an Airbus Perlan 2 glider established a world height record for an engineless aircraft, soaring to 52,172 feet over Patagonia.
It was a major milestone for the team that launched the glider, and a step toward their long-term goal of 90,000 feet. But for Simone D’Amico and his team of researchers at Stanford Engineering, it was a potentially even more consequential opportunity: It allowed them to confirm the capabilities of a virtual reality toolkit they built around a special “signal simulator” that mimicked the glider’s flight before launch.
These signal simulators are used in a variety of applications, including testing and monitoring the Global Navigation Satellite System (GNSS), constellations of orbiting satellites that pinpoint location. The devices can duplicate broadcast signals from GNSS satellites and incoming data received by sensors fitted to gliders, airplanes or satellite swarms, simultaneously plotting various trajectory scenarios for each craft.
Verifying and calibrating spaceborne navigation sensors before flight is critical, especially for the kind of precisely choreographed satellite missions required to enable swarms of satellites to provide more accurate 3D pictures of celestial or terrestrial objects. For D’Amico’s team, the Perlan 2 was a particularly good opportunity to test the simulator’s ability to confirm altitude record and establishing the feasibility of creating realistic flight plans for CubeSats, tiny but sophisticated satellites nicknamed for their shape and inspired by the smaller, cheaper, faster ethos of the computer industry.
The project got started in June 2016, when the Airbus Perlan 2 team contacted D’Amico’s group for help in planning the glider’s flight plan and verifying its altitude – an exceedingly tricky thing to do with altimeters and other standard devices because these instruments must be precisely tuned to ambient air pressure, and such calibrations become notoriously imprecise as the atmosphere thins at higher altitudes. GNSS, by contrast, doesn’t suffer from such limitations, said Todd Walter, a senior research engineer in aeronautics and astronautics at Stanford.
So, in July 2016, the Stanford team designed a computer program that reproduced the intended glider flight down to every yaw and pitch. More than one year later, soaring in the air like a condor over Patagonia, the glider duplicated the simulated flight exactly. In October 2017, D’Amico and his colleagues confirmed that the data generated in the simulator and the data received during the physical flight were absolutely congruent, establishing proof of concept for CubeSat swarm maneuvers that might be planned for future missions.
Developing the software to use the signal simulator as a tool to plan future flights promises a decrease in space mission failures, saving time, money and lives, in the case of missions involving human crews, says Vince Giralo, a PhD candidate in aeronautics and astronautics at Stanford. In 2005, he says, a NASA-sponsored mission that attempted to demonstrate autonomous navigation capabilities ended prematurely when a GNSS failure led to a collision when the spacecraft were about 200 meters apart. “If that mission had been run through rigorous testing, the error would’ve been discovered and fixes would’ve been made prior to launch, similar to our approach for future small satellite missions.”
The next big test for the lab’s new navigation and virtual reality capabilities is in 2019, when NASA will send four CubeSats into orbit as part of the so-called Starling swarm mission. The mission will be mimicked in its entirety prior to launch, says D’Amico, and should confirm the practicality of autonomous navigation for CubeSat swarms and space flight mission simulations. Moreover, he says, the simulation techniques developed by his team could – indeed, should – become standard for all space missions.
“By running these pre-flight virtual reality simulations on the ground, we greatly enhance our chances for success in space,” D’Amico says. “Things move really fast up there at several miles per second, and there’s very little if any room for error. In the simulator, we can pick apart every piece of the mission and make sure our algorithms and equipment function as expected in all orbit scenarios.”