Long before we conceived of rockets to take science into space, humans explored our own world in a much more gentle fashion.
Today, balloons are largely relegated to party decorations and state-fair rides, but once hot air balloons were the cutting edge of transport technology, able to convey people and things to places humans had never ventured before. Now, scientists are turning to balloons once more — this time, as a way to create floating space observatories and internet service stations.
Compared to the advanced rocketry and satellites favored by Elon Musk and others, balloon technology is beautiful in its simplicity. Using pressurized gas to power their journey, these balloons can soar thousands of miles above the ground into the Earth's stratosphere. But these balloon designs are not without their obstacles.
In the pair of new papers, scientists think they've solved two of the biggest hurdles to these technologies' take off — one is to do with how they are used, while the other is to do with keeping them usable.
Staying still — For balloons stationed above Earth, one of the biggest problems blocking their use is keeping the balloons "anchored," or at least hovering, above a single location. This question of how to accurately and efficiently "station-keep" was what researchers set out to answer in a new paper published Wednesday in Nature.
Trying to manually correct the course of a wind-blown balloon hundreds of feet in the air is not only a tedious task, but may also be impossible depending on the range. The balloons at the center of the paper are Project Loon balloons — these provide internet to remote areas, and are under the same parent company as Google, Alphabet.
It would be a big problem if you connected to Loon WiFi to watch a movie and the next minute the balloon was thousands of miles away. So Loon balloons have relied on a self-navigating method called StationSeek to find wind streams which will return the balloon to its station when it goes off course. This approach has worked fairly well, but is far from efficient, the authors write.
In order to accurately return to its station, the balloon must actively test and explore wind stream options, which uses more of the balloon's limited battery power.
Instead, the researchers propose a reinforcement-learning algorithm to enable the balloon to intelligently choose an optimal sequence of wind streams to return to its station without wasting excess power.
They trained the algorithm on both historic wind pattern data and nonsense "noise" to enable it to robustly make these decisions even in an unpredictable environment. This flexibility is key for the success of both terrestrial and extraterrestrial missions, the authors write:
"[S]tation-keeping offers an example of a fundamentally continual and dynamic activity, one in which ongoing intelligent behavior is a consequence of interacting with a chaotic outside world. By reacting to its environment instead of imposing a model upon it, the reinforcement-learning controller gains a flexibility that enables it to continue to perform well over time."
The method fixes the balloons above Earth, but there is no barrier to the same technology being used to fix a balloon above the surface of another world. Intelligent algorithms like this may help astrometeorologists and astrobiologists explore strange new worlds, like the atmosphere of Venus, in search of life.
A better telescope — Making balloon-led space exploration ore possible is also at the center of the second study, published Tuesday in the journal Review of Scientific Instruments. In this study, researchers from NASA's Goddard Spaceflight Center solve a very cool problem.
Space telescopes generally fall under two categories: land-based observatories and space-based observatories. But scientists want to create a third kind of telescope — a Goldilocks observatory which would hover in the liminal expanse between the Earth's stratosphere and outer space. And to hoist these living-room sized telescopes into space, scientists think giant, pressurized balloons may be the best transport.
But to work, this Goldilocks telescope need to be super cold. And to be super cold, they also need extremely heavy equipment. That means the balloon has to lift this kit up, too, Alan Kogut, a researcher at Goddard and first author on the study, explains in a statement.
"Liquid helium can easily cool the telescope, but keeping it cold means putting the entire telescope into a giant thermos bottle called a dewar," he says.
"A thermos bottle the size of a living room would weigh several tons — more than even the largest balloons can carry."
In the study, the researchers solve this problem by designing an ultra-light cooling system called BOBCAT (Balloon-Borne Cryogenic Telescope Testbed).
"BOBCAT develops technology for ultralight dewars to reduce their weight enough to allow really big ones to fly on a balloon," Kogut says. "The storage tanks are small and don't weigh much."
BOBCAT has walls as thin as a soda can and can be launched at room temperature. It cools once it reaches 130,000 feet — the Goldilocks zone scientists are aiming for.
A modified version of BOBCAT flew in 2019 to test several of the designs systems, but the team is yet to launch the ultralight system in its entirely.
When they perfect the system, Kogut and his colleagues say it will help scientists peer further back into space and time.
"Now, we have a cold telescope above the atmosphere, able to see faint images from the cold or distant universe," Kogut says.
Combining this capability with space-bound, intelligent balloons like those demonstrated by Project Loon, the future of space exploration may be a little lighter.
Abstract 1: Efficiently navigating a superpressure balloon in the stratosphere requires the integration of a multitude of cues, such as wind speed and solar elevation, and the process is complicated by forecast errors and sparse wind measurements. Coupled with the need to make decisions in real time, these factors rule out the use of conventional control techniques. Here we describe the use of reinforcement learning to create a high-performing fight controller. Our algorithm uses data augmentation and a self-correcting design to overcome the key technical challenge of reinforcement learning from imperfect data, which has proved to be a major obstacle to its application to physical systems. We deployed our controller to station Loon superpressure balloons at multiple locations across the globe, including a 39-day controlled experiment over the Pacific Ocean. Analyses show that the controller outperforms Loon’s previous algorithm and is robust to the natural diversity in stratospheric winds. These results demonstrate that reinforcement learning is an effective solution to real-world autonomous control problems in whichneither conventional methods nor human intervention suffice, offering clues about what may be needed to create artificially intelligent agents that continuously interact with real, dynamic environments.
Abstract 2: The Balloon-Borne Cryogenic Telescope Testbed (BOBCAT) is a stratospheric balloon payload to develop technology for a future cryogenic suborbital observatory. A series of flights are intended to establish ultra-light dewar performance and open-aperture observing techniques for large (3-5 meter diameter) cryogenic telescopes at infrared wavelengths. An initial flight in 2019 demonstrated bulk transfer of liquid nitrogen and liquid helium at stratospheric altitudes. An 827 kg payload carried 14 liters of liquid nitrogen (LN2) and 268 liters of liquid helium (LHe) in pressurized storage dewars to an altitude of 39.7 km. Once at float altitude, liquid nitrogen transfer cooled a separate, unpressurized bucket dewar to a temperature of 65 K, followed by the transfer of 32 liters of liquid helium from the storage dewar into the bucket dewar. Calorimetric tests measured the total heat leak to the LHe bath within bucket dewar. A subsequent flight will replace the receiving bucket dewar with an ultra-light dewar of similar size to compare the performance of the ultra-light design to conventional superinsulated dewars.