Are science-loving robots the future in space exploration?
In the short science fiction novel by Becky Chambers, A Psalm for the Wild-Built, Robots leave the factories to dwell in forests on the colony moon Panga, while humans take over all production in society. The robots spend most of their time studying nature, some go as far as to spend all of their time simply watching stalactites form or saplings grow into trees. This reminds me of the joke made by Mars researchers that Mars is a planet entirely populated by robots. It could be argued that the robotic rovers of Mars are not doing something drastically different from the robots of Panga on Mars. Rather than watching stalactites form though, they are scanning rocks for signs of chemical alteration that reveal a warmer wetter past on Mars and, most recently, what could be a biosignature.
The increasing use of artificial intelligence and robotics in society has become controversial because of concern that robots and artificial intelligence will be used to replace humans and may even turn on their human creators. It could be argued though that robots and AIs are being used properly when they are being used to extend or supplement human labor rather than replace it entirely. This is certainly true in the space environment where, at least for now, humans cannot venture far.
Automation and artificial intelligence have been used for decades for planetary exploration, dating back to the earliest lander. The Viking landers which landed on Mars in 1976 were primitive robotic landers based on pre-programmed automation. Everything the Viking landers did was pre-programmed on Earth and directly instructed to them. These instructions included the famous experiment that may or may not have indicated the presence of life on the red planet.
Since the Viking lander, an evolution of robotic intelligence in space, in a curious parallel to the evolution of biological intelligence on Earth, can be traced through the decades. The “intelligence” of the Viking landers, which is not not artificial intelligence in the modern sense, could be compared to that of a bacterium or perhaps a sea anemone or similar immobile lifeform.
The Sojourner Rover and the Mars Exploration Rovers of the 1990s and 2000s were more advanced. Because they were rovers that would be navigating terrain, they naturally had to be smarter in order to avoid obstacles. Their intelligence was based on primitive machine learning algorithms that allowed them to perform simple self-automated actions without direct commands from Earth to avoid obstacles and plan routes.
Early Mars rovers were analogous to insects or simpler animals. They had repertoire of programmed behaviors which they could perform without prior instructions in order effectively traverse an environment. They were more adaptive, being able to apply the same behavior in response to a variety of situations, but they did not learn new behaviors. The sojourner rover did not plan its own routes, but the algorithms used for the Mars Exploration Rovers were more advanced so that they could plan their own routes as well as analyze images.
More advanced machine learning was used for the Curiosity Rover, sent to Mars to search for signs of past habitable environments on the red planet. The rover’s AI system, Automated Exploration for Gathering Increased Science (AEGIS), used deep learning and neural networks, which work together to simulate specifically activity between neurons in the human brain, to enable the Curiosity rover to automatically choose rocks to zap with the ChemCam instrument to search for chemical signatures of past habitable environments. The Perseverance Rover, even more advanced, uses deep learning to enable to make its own decisions about the data that it collects and analyzes with its instruments and to travel hundreds of feet at a time on its own without human review or instruction.
Whereas the early rovers are comparable to insect or lizards, the Perseverance Rover is more comparable to a higher life-form, such as a bird or a primate. It is able to make its own decisions about the data it collects. It also is able to plan its own route with less reliance on mission control. Increasing automation will become more important as probes are sent farther out into the solar system and even possibly to other star systems where communication with humans becomes more difficult.
Ideas that are already being considered for outer solar system exploration, include insect-like climbing robots which could crawl or dig into the ice shell of Europa to get to its subsurface ocean and deploy a robotic submersible to explore under the ice and search for life. The submersible would also need to be mostly automated. Deep learning and neural networks combined with robotics provide a pathway to such a possibility. This possibility for automation, however, does not mean that humans don’t fit into the picture, at least in the near term.
Despite this capacity for automation, the Perseverance Rover still takes much longer to do its job than a human astronaut, so human astronauts may still play a vital role in the exploration of the solar system. For all we know, getting the samples collected by the Perseverance Rover back to Earth may require NASA astronauts to literally go to Mars and pick up the tubes themselves, that is, if a Chinese robotic sample return mission doesn’t beat them to it or that we don’t develop humanoid robots in the meantime that act as our direct emissaries on Mars and elsewhere.
Don’t get me wrong, I am all for human space exploration, but it does seem clear that advanced robots have advantages in the space environment. So, far we have had robots that are comparable to bacteria, insects, and higher animals exploring the space environment. Will we reach a point where robots with human-like intelligence will be able to travel to Mars, the asteroid belt, or the moons of Jupiter and explore it on our behalf?
Mars is interesting in that it is the only planet besides Earth that has its own satellite array, mainly to provide communication services to the multiple robots on the surface. In the absence of any native industry, commerce, or civilian communication on Mars, the Martian satellite array, or technosphere if you will, exists purely for the sake of scientific exploration by robots, like the Pangan robots.
A long-term exploration of the solar system may require similar infrastructure on other worlds. For example, if we want to really understand icy moons like Enceladus and Titan. The way to really understand these worlds and answer the questions we have about them, including the question of if there is life there, may require similar infrastructure.
While unlikely to happen now, could we one day have arrays of orbiters on multiple worlds supporting advanced robots powered by sophisticated AI-systems that are exploring the solar system and bringing back data to Earth for scientists to interpret and for humanity to marvel. This infrastructure of robots and satellite arrays may one day become the basis of a spacefaring civilization of scientific robots.
These robots could be very diverse in shape. There may be crawling robots to explore ice-filled lava tubes on the Moon, swimming robots to explore subsurface oceans on icy moons, rolling or walking robots to explore Mars, and flying robots to explore the upper atmosphere of Venus in search of the elusive biosignature phosphine. There could also be robotic probes traveling to nearby stars on solar sails. These robots would be trained with a deep desire to explore the universe and return to data to Earth, extending a human presence to Mars, Titan, Alpha Centuari, and beyond.
Thank you for reading my Substack! If you like what you read, feel free to share or subscribe and thank you again if you already have.
If you would like to support planetary science research, you can become a paid subscriber and double thanks if you already have.