Automata in space no prospect of replacing humans
(ORDO NEWS) — Not so long ago, NASA‘s InSight lander discovered a record-sized lake on Mars. The achievements of automata in the study of other planets are enormous – but, alas, they cannot be used to answer the most interesting questions for us. Is there life on Mars? What is hiding in the moon‘s kilometer long lava tubes? Only manned expeditions of the coming decades can tell us about all this. Let’s try to understand in detail why automata alone cannot solve the main problems of studying the nearest celestial bodies.
Machine gun or man?
The debate about who can better conquer space – a man or a robot – began in the middle of the 20th century. The designer Babakin, who headed the Lavochkin Design Bureau (the creator of the lunar rovers), was perhaps the first to clearly formulate the idea: automata can make everything that humanity really needs in space cheaper than humans. And although Korolev was a consistent opponent of this point of view, due to well-known events, in the study of other celestial bodies, the USSR, unlike the United States, had to rely only on automata.
Proponents of automata usually cite the following obvious advantages: robotic systems do not need air, water and food, so they are compact. A typical rover launches on a rocket an order of magnitude less powerful than the one needed to send humans to Mars. Accordingly, it is much cheaper to study everything with automatic machines. However, as is often the case, the road to “automatic space” was smooth only on paper. And if you think about it, it should have been expected from the very beginning.
Martian InSight: drilling no deeper than a children’s sandbox
A non-self-propelled lander InSight successfully landed on the surface of Mars on November 26, 2018. Weighing in 350 kilograms, the peak power of the device was 600 watts, which makes it a record-breaking powerful one: no other device that earthlings could deliver to another planet had such a peak electrical power. With manipulators, he put a seismometer on the surface of the Red Planet, and on February 28, 2019, he tried to start drilling.
Again, it was planned to be a record one. Until now, the deepest drilling on the Martian surface was at the level of only 7.5 centimeters (and this figure may be a little exaggerated). The Curiosity rover’s drill bit, sometimes incorrectly called a drill, simply couldn’t drill deeper. InSight was supposed to go five meters deep. This is very important, because, according to modern concepts, primitive single-celled life can exist only deeper than two meters on modern Mars: the soil at such a depth protects well from the cold and can contain a lot of water.
InSight itself didn’t have life detection instruments, they simply didn’t fit. During drilling, only the heat flux from the depths of the planet was to be measured to clarify the geological activity of its interior. But the very first test of deep drilling on Mars would at least show that an automatic search for life is possible here.
Alas, the drill could only go 35 centimeters deep before getting up for some unknown reason. He either bumped into a stone, or he simply could not work normally in crumbling soil. Although this is still a record deep drilling, the device did not achieve its goal. He could measure the heat from the bowels of the planet at a depth of no more than three meters. InSight’s core mission has so far been thwarted, and it is doubtful it will ever be accomplished. A $ 830 million mission doesn’t often end so sadly.
The device is limited in weight, so it has no mobility (it would take hundreds of kilograms on the chassis). He cannot move the drill to another place. Perhaps this is one of the best illustrations of the limited capabilities of automata on other celestial bodies. A person with an elementary spade could go deeper, and if there is a solid rock there – move a little to the side. Alas, that is easy for us, while it is not available to our machines on other planets.
At the same time, InSight received a large consolation prize. According to preliminary data leaked to the press, he discovered a layer of electrically conductive material four kilometers thick under the surface of Mars in the landing area with magnetometric instruments. The Red Planet does not have active geology, and the only candidate proposed to date for filling such an electrically conductive layer is water with salts dissolved in it.
If this data is confirmed, it turns out that under the InSight lies a fairly deep subsurface sea. This is the name deserves a reservoir with a water column of four kilometers. Such reservoirs have long been considered one of the best candidates for habitats of the simplest life. Similar places in Antarctica (Lake Vostok) contain microorganisms, despite the fact that they have been hidden under the ice for many millions of years.
For a separate explanation: InSight is not really a robot. He drilled (at least tried) not according to a given program, but according to a series of commands that were given to him from Earth.
Where are the autonomous robots?
The question may arise: why are we talking about remotely controlled machines, and not about autonomous robots? A typical production drone for about $ 500 has both cameras and software to avoid trees, branches and people in flight. On more serious wheeled drones from Yandex or Waymo, there are more cameras and even lidars, and the software is more serious, so they see obstacles in a complex environment hundreds of meters ahead. Why can’t you do something like this for a rover?
Unfortunately, the wave of news about how much modern unmanned aerial vehicles (UAVs) can do, largely obeys the “unmanned” fashion, which explains their capabilities somewhat exaggerated. The press rarely reminds us that UAVs crash 30-300 times more often per hour of flight than conventional aircraft. Moreover, the main causes of crashes are technological, and not at all operator errors. The most common cause of a UAV accident is the banal loss of communication between the drone and the operator: the drone switches to a fully autonomous mode, and the latter is just inferior to the human operator in the ability to fly without an accident.
The large accident rate of UAVs is not accidental, but natural: there is no strong AI today, and without it, a drone cannot have the same safety of movement as a person with his brain. To control the movement, you need to understand what you see in front of you: for example, what obstacles lie in the way and whether they are. The current AI can effectively distinguish objects (branches, wires, other obstacles) in favorable conditions when they correspond to the images embedded in its memory.
But the glint of the sun, the shadow of a tree, and a host of other factors can make the seemingly cataloged image of the obstacle invisible to the drone. What we call vision, in fact, is inseparable from the ability to recognize images. A sparrow does it very well, but a drone is not a sparrow: its “brains” will not allow flying in difficult conditions without an accident for a long enough time.
It is for this reason that the autopilots on modern aircraft are not at all “robotic flight”, but only advanced cruise control, which makes the work of pilots easier. Airplane autopilot is similarly devoid of intelligence, so pilots are depressed and can use it to crash the airliner they control uphill. And, as noted by the pilots of airliners, it is difficult not only for victims of depression: the key problem of the autopilot is recognized as “the inability to make non-standard decisions that depend on the specific situation.”
It’s actually much easier for flying drones to do without a human operator than on ground ones. Air is a homogeneous medium where the density of obstacles is very low, there is “a lot of space” in it and there are not many objects. The terrestrial environment is flat and densely littered with large obstacles that must be noticed and avoided. In addition, the surface with which the chassis of the ground robot is coupled is extremely heterogeneous: there are holes, stones, sand, and much more.
Therefore, in fact, there are no “self-driving cars” about which the press writes so much, either. Newspapers are happy to put the phrases “Waymo launched an autonomous taxi service in Phoenix” in the headlines, but rarely mention the facts about such services. And they consist in the fact that in the “unmanned taxi” in Phoenix there is exactly the same engineer who must take control in case of any unforeseen situation.
Even as such, this taxi is not suitable outside of the high-resolution lidar mapping zones. When these maps become outdated – when the road is being repaired and a new traffic light is being installed – the lidar drone simply drives at a red light, as it did on the streets of San Francisco. To complicate matters further, lidars in the air with water vapor do not work well, so Waymo launched its test service – by the way, inaccessible to ordinary taxi passengers, only to a limited circle of people – in the capital of the not-rainy state of Nevada.
For marketing purposes, both Yandex and Waymo can talk long and hard about their “self-driving cars”, but in practice they do not work without a person behind the wheel outside of promotional trips. And if real drones-cars appear, then only those developers who are engaged in automated imitation of human drivers, but not where they are trying to drive based on standard algorithmic solutions developed without the participation of a truly huge training sample based on human driving.
The problem is that there is currently no sample of millions of people driving on Martian off-road. Algorithmic solutions without this will not give real, relatively accident-free drones on Earth, so it is doubly difficult to talk about their use on Mars. Even if they suddenly appeared, we are talking about difficult solutions with high energy needs.
The Nvidia Drive PX Pegasus delivers up to 310 trillion operations per second while consuming 500 watts. This is several times more than the entire available constant power of the Curiosity rover (a little over a hundred watts) – and indeed any rover in history. And we have not yet calculated the power of the lidar, many cameras, ultrasonic sensors. Of course, there are more energy efficient solutions (a Tesla computer requires only 72 watts and does not need a lidar), but they can work, recall, only through training on a huge sample of real driving human drivers. Simply put, this cannot be realized on the Red Planet.
The only chance for the autonomy of the apparatus for exploring other planets is flying in the atmosphere, where there are no boulders, the danger of slipping and many other troubles. But the Martian atmosphere roughly corresponds to the Earth’s low vacuum: flying there is energy-intensive. Venus is generally not suitable for flying close to the surface – it’s hot. The moon, asteroids and large satellites such as Europa and Ganymede are atmosphere less, and it will not work there.
Theoretically, Titan is suitable, where NASA plans to send the Dragonfly. It is 38 times less energy-intensive to fly there than on Earth (the atmosphere is four times denser). In practice, the route selection and actions of a billion dollar device are much more reliable to do with the help of a remote human operator. If UAVs on Earth fall more often than manned vehicles, then this is just another lost UAV. On Titan, it will be a loss of a billion dollars and many years of effort by NASA’s top talent. No one will take such a risk – it just doesn’t make sense.
In other words, any mobile devices for studying the surface of other planets will be de facto remotely controlled in the entire foreseeable future – fortunately, there are no clear prospects for creating a strong AI today.
Why NASA “Dragonfly”
In June 2019, NASA announced its intention to send a Dragonfly to Titan. It is enough to look at the external appearance of this eight-rotor flying machine to understand how it differs from the classic appearance of the rover, which has basically remained unchanged since the days of the “Lunokhod”. It has no wheels, only a helicopter-type landing gear. In order to move, he must fly.
As you know, the flight is energetically expensive. This is especially true because a small device does not have much energy: it will be powered by radioisotope thermoelectric generators with an efficiency of only a few percent. This means that he will need a noticeable amount of plutonium-238. To feel the urgency of the situation, it is worth remembering: such a generator gives only 100-125 watts of power with a weight of 45 kilograms. Curiosity is powered by such a system on Russian plutonium-238, but it is enough: six wheels are rotated by a small electric motor. For flight, more energy is needed.
But the aircraft cannot be too heavy (the mass of the Dragonfly is only 300 kilograms), that is, it will have to take more fuel and less scientific instruments. Maybe that’s why there is no drill on board, although it would be good to know something about the bowels of Titan. Why does NASA make such sacrifices, forcing to fly what, it would seem, was born to crawl?
Everything is explained simply: in flight there is no danger of getting stuck in a dune – navigation in the air is easier than on the surface, and is quite accessible to software. Therefore, the Dragonfly, according to the plan, will be able to cover as much as 175 kilometers during the service – and at the same time look into a dozen different places near the equator of Titan. As NASA proudly writes, this is “roughly double the distance covered by all rovers combined.” Perhaps this is the key phrase.
Why rovers are so ineffective
It is not for nothing that NASA estimates the 175 km range as an achievement of enormous significance. To study an alien celestial body, you need to see its different parts. In some places on Mars, methane is emitted, in others it is not. Some have something like streams of water, while others do not. Some interesting regions are at a great distance from others. Getting to more than one of them during a robotic mission would be important.
Meanwhile, the automata do it very badly. Lunokhod-2 traveled more than 300 meters a day. Opportunity covered about three kilometers in a year. The reason for such agility of the ancient Soviet machine is not that Americans in the 21st century make bad rovers, but that the very term “automaton” in relation to space exploration is of little use, rather misleading fans of the concept that a person has nothing to do in space.
To be more precise, there are no automatic planetary rovers today. Like the Moon Rovers, American Mars rovers are remotely controlled machines, something like the ones that each of us used to have fun with as children. As with toy cars, they must be handled carefully on rough terrain, otherwise they will skid and stand. This is how Lunokhod-2 perished.
Likewise, Opportunity was almost fatally stuck in 2005, and the “dune” in which he did it was only 30 centimeters high. A grown man can push a car out of such a “dune” by hand, but there is no man on Mars. And if not for luck and long attempts to get out, the rover could have stopped at the beginning of the journey.
The reasons why Opportunity traveled less in a month than Lunokhod-2 in a day lie in the same area. The remotely controlled vehicle is 1.3 light-seconds away on the Moon and at least 280 light-seconds away from the Earth operator on Mars. More importantly, the Moon is always turned towards the Earth on one side, so communication with the lunar rovers was constant. Mars rotates, and communication sessions (less than an hour) happen once (maximum two) per day.
At first glance, it seems that communication can be made permanent by simply placing an analogue of a geostationary relay satellite in the areocentric orbit of Mars (17 thousand kilometers above it). In practice, a dedicated satellite in such an orbit will be heavy (the areostationary orbit is located between the orbits of Phobos and Deimos, which will cause the satellite to spend a lot of fuel on frequent orbit corrections).
And no one will provide a rover with such an expensive and complex partner as an areostationary satellite, the life of which, judging by its geostationary colleagues, may be noticeably shorter than that of the rover itself. The same Curiosity still uses relay satellites, because the power of its transmitter to the Earth directly could not be enough. But these satellites pass over the apparatus in tens of minutes, the same amount of sessions of communication with the Earth. It is difficult to call such a permanent connection.
Seeing an interesting stone during the next communication session, the rover operator instructs him to drive a little towards this stone. And the next day he looks to see if there is a sand dune on the way to the desired stone, 30 centimeters high, which is deadly for a machine gun. If it is not there, gives the command to drive a little more. On the third day, if the stone was close, the command is given to stretch the manipulator to the stone. On the fourth, they look to see if the manipulator took the stone correctly. If yes, they give a command to work with him. And so on, without end.
Only work where it is easy
It may seem that the sluggishness of the “automata” (which in fact are not automata at all) can be neglected. Though slowly, they will do their job. Alas, everything is a little more complicated. Any rover today is doomed to work not where scientists would like to, but where it is easier for him. No one will chase him down a steep slope with loose soil – like the one on which traces of streams suspiciously resembling water were found. The billion-dollar device avoids even 30 centimeters of sand on a flat surface – it has no time for steep slopes.
But this is just the tip of the iceberg. The most interesting places of the Red Planet and the same Moon with Titan are located under their surfaces. Here we are not even talking about the huge underground lakes of Mars or Titan’s hypothetical subsurface ocean. We are talking about lava tubes very close to the surface. These are huge, many kilometers (due to low gravity) cavities left by hot gases during volcanic eruptions on the Moon and Mars.
Both there and there in the past there was a lot of water and water vapor on the surface, which is why there are suggestions that a lot of ice has accumulated in lava caves – both water and, possibly, carbon dioxide. Lava tubes can theoretically go down quite deep – to where the temperature is above zero degrees Celsius. In Mars conditions, this could mean liquid water – and potential life. It would be nice to explore lava tubes, especially since their outlets there are often open and large.
This is not possible with machine guns. You cannot send a radio-controlled car into a cave where radio signals cannot pass. Even the repeater at the entrance (which will take away additional mass and energy) will not help once the rover turns the corner. And will he be able to move there? Smooth ice at the entrance can be overwhelming for a car stuck in 30 centimeters of sand. The slopes of the lava tubes are quite steep: you can go down there on a cable, but automata are not masters of mountaineering, even on Earth, let alone on Mars or the Moon.
Unfortunately, there is more to come. An automaton cannot be human-like in terms of the versatility of manipulators and skills. A person is able to dig a rather deep hole with an entrenching tool and even reach a depth of more than a meter with a hand drill. An automatic drill can only penetrate so deeply if it is a meter-sized device with a large mass. It’s too heavy for a rover.
Therefore, in practice, they have small, “pocket” drills, and the drilling record set by the rover for today is 7.5 centimeters (Curiosity). And all the most interesting things on Mars lie deeper than a meter: it is there that the soil effectively protects against radiation and temperature changes on the surface.
At the same depth, bacterial colonies exist in terrestrial deserts. In 2011, Spanish and Chilean explorers simulated the search for life on Mars in the Atacama Desert. It was chosen for the content of chlorides and perchlorates in the soil – components typical of the Martian regolith. As a result, when analyzing samples recovered by machine drilling, they were able to find life there, but at a depth of at least two meters. This means that even if Curiosity had traveled directly over such bacterial colonies for all seven years of its operation, it would not have been able to detect them. It was easy for researchers to deliver a large drill to Atacama, but it is difficult to throw it on Mars.
Another important point: bacterial colonies are scattered in spots under the surface of the Atacama. In ground tests, it is easy to move the five-meter drill from place to place. On Mars, such a device is too heavy for mobile rovers. And a stationary lander with a drill may simply not be lucky to stumble upon a local analogue of the Atakam bacterial colony. In other words, the devices that we send to look for the simplest life on Mars are technically unable to find life even in the terrestrial desert.
Of course, you can send a specialized stationary automatic driller to the Red Planet: suddenly he gets lucky, and something interesting will accidentally be right under his landing point. This is exactly what NASA did by sending InSight to Mars. True, this particular device is not looking for life, but, as will become clear below, the problems of drilling have not escaped it either. Unfortunately, this “automatic machine” (all its actions are controlled by radio commands from the Earth) has not yet shown its best side. Recently, he either lost traction with the ground, or stumbled on a stone – as a result, the drilling stalled in the first 30 centimeters.
A number of media outlets soon wrote optimistically: “InSight moved the HP3 drilling rig to a different location” – in theory, this allows you to breathe a sigh of relief. In fact, this is not the case: the machines are highly specialized. InSight simply cannot move the rig: it is too large for this, and the capabilities of the “automatic” machine are too modest. He only moved the supporting structure for his drilling device to another place so that the operators could better see what was happening to him.
If there is a separate large stone under it, drilling will not work further. Because this “machine” is so specialized that, as NASA reports, “the robotic arm is not designed to lift a“ mole ”(HP3 drilling device. – Ed.) After it has been separated from the supporting structure, so that it is not will be able to rearrange the “mole” if the rock blocks his way down. ” So, we are once again convinced: remote-controlled machines are narrow specialists, whose manipulators do not have the same wide capabilities as human hands. The latter can lift the load, but the InSight “arms” cannot.
In theory, a large launch vehicle can deliver a lander to another planet, from which it will launch a large mobile drilling complex weighing tons, capable of both drilling meters in depth and moving between drilling at the required distance. And even attach a take-off module to it to return hundreds of kilograms of soil to Earth. But one must understand that the cost of such a remotely controlled vehicle will, if necessary, approach a manned expedition: to softly land several tons on another planet means the use of landing gear, in complexity and weight comparable to a landing module for humans.
The mass of a pair of astronauts with everything necessary for a short landing (according to the experience of 1969) is comparable to the weight of the same Curiosity rover. At the same time, a manned expedition has a lower chance of fatal stuck when moving (this is how submachine guns died on the Moon and Mars) and will still be noticeably more versatile than a specialized mobile drill.
All this means that while the machines for studying the planets “in place” work like a character in a joke who lost his keys at a dark entrance, but is looking for them under a lantern – only because it is brighter there. They don’t explore the depths that are most promising. They choose as flat an area as possible, without dunes, where they can work somehow.
Whatever one may say, to find out in this way whether there is life on Mars, as well as to explore the most interesting places of the Moon, will not work. Obviously, if we humans want to answer these questions, then all that remains for us is to prepare manned expeditions there. Fortunately, in recent years, much has been done in this direction. If not in the 2020s, then in the 30s of this century, a person can still reach the Red Planet. And there the answers to the cherished questions will be found – and sooner rather than later.
—
Online:
Contact us: [email protected]
Our Standards, Terms of Use: Standard Terms And Conditions.
