Autonomous Spacecraft Are the Next Step for NASA and the ESA
Both NASA and the European Space Agency (ESA) have a lot of ambitious plans on the horizon, yet none of these are feasible without autonomous spacecraft that can “think” on their own.
While autonomous vehicles are hailed as the future of transportation on Earth, astrophysicists have also been quietly developing spacecraft that can respond to their changing and unpredictable surroundings without the intervention of humans. Once deployed, these types of spacecraft will complete complex tasks while handling unforeseen complications on the surface of a distant planet, hurtling through deep space, orbiting the Earth, among others.
NASA and the ESA hope to employ autonomous spacecraft as a planetary defense shield, to explore Mars, to begin building a moon colony, and to avert the Kessler syndrome. All of these are exceedingly difficult to do with humans due to safety concerns, such as radiation exposure, and the need for food, water, oxygen, and other supplies. Humans also require almost constant communication with ground control, and these complex future missions would require so much data being sent back and forth that it would surpass the ability of current technology. NASA scientists claim this is from “the latency and bandwidth constraints on communications between the vehicle and ground control.”
For example, it takes NASA’s Voyager 1 around 20 hours to send a signal back to Earth from beyond the edge of the solar system, which it reached in 2012. Therefore, the next generation of NASA and ESA spacecraft need to function on their own, without humans on board and with limited communication with the ground.
“The spacecraft that support these challenging future missions will need to be capable of reasoning about their own state and the state of their environment in order to predict and avoid hazardous conditions, recover from internal failures, and meet critical science objectives in the presence of substantial uncertainties.”
Jet Propulsion Laboratory
How Do Autonomous Spacecraft Work?
Like self-driving cars, autonomous spacecraft detect their surroundings mainly by bouncing EM waves off nearby objects. For example, radar uses low frequency EM waves, which are beneficial in the presence of dust, rain, snow, gas, etc., as they simply pass through. Lidar on the other hand uses high frequency EM waves, which are reflected easily by small particles, although it offers much more detail. Autonomous vehicles on Earth use these to detect other cars, buildings, a person crossing the street, etc. Autonomous spacecraft can use these to detect their distance from Earth, asteroids of all sizes, other spacecraft, geological features in an alien landscape, etc.
Furthermore, on Earth, autonomous vehicles use GPS to determine its location and the location of other objects, but this is not possible in space. Some scientists, though, have proposed the idea of using pulsars in much the same way GPS uses its network of satellites. Pulsars are neutron stars that emit high intensity EM radiation from its poles with extreme regularity. In fact, some pulsars that have rotations in the milliseconds are more precise than atomic clocks. These jets of radiation can be detected from Earth, much like the light from a lighthouse. The distance to several pulsars can determine an object’s exact location, just as the distance to satellites can be used to pinpoint something on Earth. In 2018, NASA showed this was possible with an experiment on board the International Space Station called the Station Explorer for X-ray Timing and Navigation Technology. They claim it “showed that millisecond pulsars could be used to accurately determine the location of an object moving at thousands of miles per hour in space.” Naturally, this will be essential for autonomous spacecraft to navigate the depths of space.
For example, Pioneer 10 and 11, NASA spacecraft launched in 1972 and 1973 respectively, featured a plaque that pinpointed the location of Earth by showing its distance to a handful of pulsars. It was thought that any intelligent alien species would be well aware of such powerful and regularly rotating objects and could use them to find our home planet.
Autonomous spacecraft will also rely on high-resolution photography, just as autonomous vehicles on Earth do. These are fed into various machine learning algorithms that have been trained to recognize different objects and predict how they are moving. For example, the Scale Invariant Feature Transform is an algorithm that takes an image, breaks down its key features, and compares it to a database of already understood objects. If it sees a picture of a stop sign, it will recognize its eight vertices and conclude it must be a stop sign. Other algorithms include the Histogram of Oriented Gradients, TextonBoost, You Only Look Once, among many others. All of these take a unique approach to identifying objects and their movements, while algorithms like AdaBoost help different algorithms work together, thus minimizing their weaknesses. With these types of algorithms, autonomous spacecraft can use a series of photographs to understand their surroundings and make predictions about how their situation will change, such as recognizing and predicting the path of an asteroid.
Gyroscopes, accelerators, thermal cameras, star trackers, and many other types of detectors will be used as well to produce as much data as possible. All of this data is fed into the “brain,” which trains itself through successive generations of trial and error. By using what is known as deep learning, a neural network can refine itself to find the optimal solution to tasks that are insurmountable to humans. Autonomous spacecraft can train themselves on countless simulations, honing their understanding of a situation before engaging in the real thing.
For example, Google’s AlphaZero used deep learning to beat Stockfish, the world’s strongest chess engine, in about 4 hours, and now is untouchable by even the best in the world. Likewise, AlphaStar is dominating the world of StarCraft, and AlphaGo is the unquestionable world champion in the game of Go. Although these are just games, it shows that deep learning in the right hands can produce “brains” which can understand and master complex tasks. Go has an estimated 10^360 possible games, chess has an estimated 10^120. For comparison, there are only about 10^80 particles in the observable universe.
Furthermore, part of the computations for an autonomous spacecraft may be done on blockchain. The smart contract feature on the Ethereum blockchain can be used to “develop a decentralized, secure, and cognitive networking and computing infrastructure for deep space exploration,” according to Wei Kocsis. The Ethereum blockchain is almost Turing complete, meaning it is near computationally universal, which allows it to perform calculations and complete computational problems across several nodes. NASA awarded her over $300,000 to research how this new technology can be integrated.
NASA’s Jet Propulsion Laboratory wants their autonomous spacecraft capabilities to “include automated planning, intelligent data understanding, execution of robust critical activities such as entry, descent and landing (EDL), and situational- and self-awareness.”
Planetary Defense System
Hera is an ESA spacecraft that will head to an asteriod called Didymos, which has its own moon named Didymoon. When they wiz by Earth in 2022, Hera will launch soon after to make contact with the pair and collect extensive data about them. Ultimately, Hera’s mission is to run an “asteroid mitigation precursor experiment whose objective is to change the orbit of Didymoon.” This will not change the pair’s path as they go around the sun, but the resulting change in Didymoon’s orbit will provide crucial insight into the effectiveness of different techniques to change the path of an orbiting body, information that will certainly be useful if/when an Earth threatening asteroid is detected.
This mission will also collect critical data on Didymos and Didymoon’s internal structure and mineral composition.
More importantly, this mission will test its ability to take information from its main sensor, the Asteroid Framing Camera, combine it with data from its many other sensors, and make real time decisions. The Asteriod Framing Camera has seven spectral filters, allowing it to gather immense amounts of data in a broad range of the EM spectrum. If it goes well, this is a major step towards an autonomous planetary defense system against asteroids, especially ones that may cause another mass extinction.
In-space Manufacturing
Autonomous spacecraft can also be used for building other spacecraft while in orbit or on an extraterrestrial body, such as a moon colony. NASA has recently given $142 million to Colorado based Maxar Technologies to build the Space Infrastructure Dexterous Robot (SPIDER), which will go aboard the Restore-L satellite sometime in the next few years. Restore-L has the mission of extending the life of satellites by helping them reposition and refuel if needed. SPIDER’s mission is to construct a 3 meter long communication antenna without the intervention of humans.
“In-space assembly and manufacturing will allow for greater mission flexibility, adaptability, and resilience.”
Brent Robertson, SPIDER project manager
Likewise, the United States Naval Academy built the Intelligent Space Assembly Robot (ISAR) to advance the field of off-Earth autonomous manufacturing. ISAR has a 3U Cubesat design, meaning it is small and lightweight, making it easy and cheap to launch and test in orbit. ISAR will have two robotic arms, each with six degrees of freedom, and numerous sensors that will provide not just visual but tactile data.
“Due to the high availability of CubeSat launch opportunities, ISAR can be launched a number of times which allows the system to rapidly advance its autonomous capabilities.”
Space Foundation
The Archinaut One will launch in 2022. Its purpose is to demonstrate the ability to 3D print two 10 meter beams, which will eventually be used to support solar panels, of course, without the intervention of humans. NASA awarded California based Made In Space nearly $75 million to make this happen. It also has a robotic arm to aid in future construction missions.
Exploring Mars
The Mars 2020 rover has a launch date of July 2020, with an estimated landing date of February 2021. Its main mission is to collect 20 samples of rock and soil and keep them in a cache for future pickup. While this is not entirely done on its own, the rover will have far more freedom than its predecessor, Curiosity. Mars 2020 will be able to roam without constantly checking in with ground control and it will be able to figure out and implement its own schedule to efficiently use electricity and other resources. NASA claims that the rover’s new brain “allows it to shift the time of some activities to take advantage of openings in the daily operations schedule.”
Furthermore, Marco Pavone and other researchers from Stanford developed a mothership named the Phobos Surveyor and several small rovers named hedgehogs, which are contained inside. If ever launched, the idea is that the hedgehogs would be deployed one at a time, allowing them to bounce across the surface, collect data, and send it back to the mothership and the other hedgehogs. Based on this information, the mothership can decide where, when, and how to continue hedgehog deployment to maximize resources, efficiently complete missions, and avert danger. The original plan was to explore the surface of Phobos, a Martian moon, but the researchers believe this could be deployed on Mars or the moon just as easily.
The Kessler Syndrome
Proposed in 1978 by Donald Kessler, this is a hypothetical scenario in which there is so much space junk in low orbit that any future space launches are impossible. Over time, abandoned satellites and other orbiting objects collide, resulting in countless pieces of debris, creating a band around the Earth that cannot be feasibly crossed, as even minor damage by a small piece can render a spacecraft unusable. While we are not close to this happening yet, some scientists are becoming concerned, as there are about 34,000 known man-made objects over 10 cm in orbit.
Removing large debris like satellites would require a separate expensive, complicated, and dangerous mission for each one, if done with a manned spacecraft. However, autonomous spacecraft would be able to solve this problem much more effectively. In fact, a Swiss startup by the name of CleanSpace is now being supported by the ESA because they have developed a reusable spacecraft that can target, capture, and dispose of space debris by sending it to burn up during reentry. This process will be done only partly autonomously, although this is a big step towards averting the Kessler Syndrome.
The Future of Space Exploration
The most ambitious off-Earth plans for the near future include exploring and sending astronauts to Mars, sending astronauts back to the moon, establishing a moon colony, exploring Europa and other moons, mining asteroids, etc. All of these will not be possible unless autonomous spacecraft are used, as they reduce the amount of communication needed with ground control, have longer mission lengths due to not having humans, can perform tasks much more efficiently than humans, can adapt to unforeseen circumstances quicker than humans, can be trained with deep learning to surpass the abilities of humans, among many other benefits.
They can even perform routine tasks without being prompted or controlled. For example, Astrobee consists of 3 robots which have already been deployed aboard the International Space Station. They work largely autonomously and can perform a range of complex tasks when they sense it is needed. For example, they can take inventory, document experiments being done by the astronauts, move cargo, etc. They use fans to propel themselves throughout the station, and they use many different types of sensors to navigate and determine if work needs to be done.
“Robots such as Astrobee, have the capacity to become caretakers for future spacecraft, working to monitor and keep systems operating smoothly while crew are away.”
NASA
Therefore, NASA, the ESA, and other space agencies around the world have been pouring billions of dollars into making them happen, the fruition of which we will begin to see shortly.
If you enjoyed the article, please consider donating!
